Spontaneous Appearance of Triiodide Covering the Topmost Layer of the Iodide Solution Interface Without Photo-Oxidation

Ions containing iodine atoms at the vapor-aqueous solution interfaces critically affect aerosol growth and atmospheric chemistry due to their complex chemical nature and multivalency. While the surface propensity of iodide ions has been intensely discussed in the context of the Hofmeister series, the stability of various ions containing iodine atoms at the vapor–water interface has been debated. Here, we combine surface-specific sum-frequency generation (SFG) vibrational spectroscopy with ab initio molecular dynamics simulations to examine the extent to which iodide ions cover the aqueous surface. The SFG probe of the free O–D stretch mode of heavy water indicates that the free O–D group density decreases drastically at the interface when the bulk NaI concentration exceeds ∼2 M. The decrease in the free O–D group density is attributed to the spontaneous appearance of triiodide that covers the topmost interface rather than to the surface adsorption of iodide. This finding demonstrates that iodide is not surface-active, yet the highly surface-active triiodide is generated spontaneously at the water–air interface, even under dark and oxygen-free conditions. Our study provides an important first step toward clarifying iodine chemistry and pathways for aerosol formation.


■ INTRODUCTION
Halide ions at the vapor-aqueous solution interfaces play an essential role in aerosol nucleation and atmospheric chemistry.Among various halide ions, iodine exhibits appreciably more complex chemical behavior by varying its oxidation number from +5 to −1 and showing the lowest electron affinity. 1odide (I − ) can be converted to triiodide (I 3 − ) and iodate (IO 3 − ), 2,3 leading to polyvalent iodine compounds. 4Polyvalent iodine compounds at aqueous interfaces play a critical role in the nucleation of aerosol particles and the emission of iodine into the atmosphere.−11 In particular, the conversion of I − into I 2 in the absence of an oxidizing agent has been discussed in the past, while it is debatable. 2,12nveiling the stability and surface propensity of various iodine derivatives at the vapor−water interface is essential for understanding the formation process and lifetime of aerosol particles.
Various theoretical and experimental techniques have been used to investigate the surface of the iodide salt solutions.Theoretically, molecular dynamics (MD) simulations have examined the surface propensity of halide ions; 13−22 ions with a large radius tend to have a significant surface propensity, while the predicted surface propensity depends highly on the accuracy of the force field models.The surface of the aqueous iodide solutions, such as NaI and KI solutions, has been probed with interface-specific spectroscopy, including sumfrequency generation (SFG), second-harmonic generation, 23−31 (photo)electron spectroscopy, 32−38 and mass spectrometry. 39However, such theoretical and experimental studies have assumed that only iodide is present at the vapor− water interfaces when an iodide salt solution is prepared, despite the many polyatomic iodine compounds that have been proposed to be present at interfaces. 3,5,40The following question thus arises: What is the iodine species that is stably present and covers the topmost layer of the NaI solution in the nitrogen and dark environment?
To address these questions, we probe the N 2 gas-D 2 O solution at the NaI interface by combining polarizationdependent heterodyne-detected SFG (PD-HD-SFG) with ab initio MDs (AIMD) simulations.The PD-HD-SFG technique 41 allows us to unambiguously quantify the density of free, non-hydrogen-bonded O−D groups at the interface by accounting for possible changes in the molecular orientation. 42,43This is achieved by measuring the imaginary part of the complex-valued second-order susceptibility (χ eff (2) ) for different polarization combinations. 44Since standard force field simulations can misrepresent the ion−water interaction, 45 AIMD simulations at the ab initio level of theory are required to simulate the free O−D behavior and the stability of ions at aqueous interfaces.
Here, we show that the free O−D population is drastically reduced with increasing NaI concentration ([NaI]) in D 2 O solution, but only when [NaI] > 2 M. AIMD simulation reveals that the drastic reduction of the free O−D population cannot arise from interfacial iodide adsorption but rather can be attributed to the adsorption of triiodide ions.We further infer that the concentration of triiodide at the interface is ∼10 8 times higher than that in the bulk.We discuss the interfacial chemical equilibrium and its implications for atmospheric chemistry.

■ RESULTS AND DISCUSSION
Figure 1a and 1b display the measured Imχ eff,ssp (2)   and Imχ eff,ppp (2)   spectra at the N 2 gas-D 2 O solution interface with varying [NaI], respectively.The Imχ eff,ssp (2)   spectrum at the N 2 gas-neat D 2 O interface shows the 2730 cm −1 positive peak, 2650 cm −1 positive shoulder peak, and 2550 cm −1 negative band, consistent with the previous study. 46The 2730 cm −1 peak, the 2650 cm −1 shoulder peak, and the 2550 cm spectra, while the peak amplitude below 2720 cm −1 decreases.
From the χ eff,ssp (2) and χ eff,ppp spectra, we obtain the yyz and zzz components of the second-order susceptibilities, χ yyz (2) and χ zzz (see Supporting Information), 51 where the xz-plane forms the incident plane and the z-axis forms the surface normal.Because we are primarily interested in quantifying the response of the free O−D groups on the surface (rather than that of the hydrogen-bonded O−D groups below the surface), the relevant Fresnel factors were calculated using the interfacial refractive index deduced from the simulations in a similar manner to refs 52−54 for the free O−D groups' interfacial location (see Supporting Information Section 3.2).The inferred Imχ yyz (2) and Imχ zzz (2) spectra are displayed in Figure 1c  and 1d, respectively.Both spectra show that the free features of the elongated O−D feature diminish with increasing [NaI].A closer look at the peak amplitude extracted from the fits (Figure 1e) reveals a highly nonlinear behavior, also apparent from the raw data: the Imχ yyz (2) and Imχ zzz (2) amplitudes of the free O−D peak (a yyz and a zzz , respectively) only start to decrease above [NaI] = ∼2 M. The surface pressure data of the NaI solution also shows a similar nonlinear behavior above ∼2 M (see Supporting Information).In contrast, we could not see such a nonlinear trend for other halide ions, such as Cl − , as displayed in Supporting Information. 50ince the SFG peak amplitude of the free O−D peak is determined not only by the surface density of the free O−D stretch chromophores (N S ) but also by the orientation of the free O−D group, 51,55 one needs to disentangle these contributions to access N S .To do so, we extracted the free O−D peak area in the Imχ yyz (2) and Imχ zzz (2) spectra (denoted as A yyz and A zzz , respectively).Subsequently, we obtained N S using the following relations 56 A N r r 1 2 (( 1) cos ( 1) cos ) A N r r (( 1) cos cos ) spectra were obtained at the N 2 gas-D 2 O interface in the absence and presence of NaI.Absolute amplitudes were obtained using the known effective surface nonlinear susceptibility of z-cut a-quartz (see Supporting Information).(c) Imχ yyz (2) and (d) Imχ zzz (2) spectra after correction for the Fresnel coefficients.The error bars from multiple measurements of these spectra are given in Figure S1.(e) Variations of the free O−D peak amplitudes, a yyz and a zzz , obtained from fits to the Imχ yyz (2)   and Imχ zzz (2) spectra, respectively (see Supporting Information).Lines serve to guide the eyes.

Environmental Science & Technology
where θ denotes the average angle of the chromophores with respect to the surface normal, α is the hyperpolarizability, and r is the depolarization ratio of the free O−D stretch mode.We used r = 0.15. 57By using an exponentially decay-shaped function for the orientational distribution 57 and assuming the rotational motion to be slow compared to vibrational energy relaxation, 43  Iodide (I − ) has long been suggested to be a species with a high surface propensity. 13,15,16,23Can the reduction of N S be rationalized by the appearance of iodide in the topmost water layer?To answer this question, we carried out AIMD simulations at the vapor-D between the experiment and simulation illustrates that the presence of iodide alone cannot explain the substantial reduction of N S observed in the experiment.In fact, the concentration profiles in Figure 2b−d show that the I − density at the interface (z ∼ 0 Å) is comparable with that in the bulk (z < −7 Å) at all concentrations.This means that the iodide is not particularly surface-active, and the reduction of the contact area of the vapor−water interface due to the iodide is limited.Note that this trend differs from the previous classical MD simulations [13][14][15][16]18 but is consistent with another AIMD simulation, 21 indicating that force field parameters may erroneously lead to an apparent stabilization of I − at the vapor−water interface. This otion is consistent with the recent finding that the repulsive force of the Lennard-Jones potential, 45 together with the underestimated dielectric property of simulated water, 58,59 erroneously predicts ions to be surface active.
Clearly, the surface activity of the iodide ion cannot explain our observations.What moiety is formed and appears at the N 2 gas−water interface, reducing the free O−D group population (N S )?A plausible candidate is a polyvalent iodine ion such as IO − , IO 2 − , IO 3 − , and I 3 − .Among these species, we can rule out IO − , IO 2 − , and IO 3 − as the surface active species since these ions are more hydrophilic than iodide and thus will be located in the bulk of the water.Indeed, a recent AIMD study 60 has shown that all the atoms of the IO 3 − ion can form strong hydrogen bonds with water molecules, implying that the surface propensity of IO 3 − ions would be limited.We also measured the Imχ yyz (2) spectra from the surface of a NaIO 3 solution and found that, indeed, the free O−D peak feature is unchanged upon the addition of NaIO 3 (see Figure S5), confirming that IO 3 − is not a species to change N S of the free O−D groups.
After excluding IO 3 − as a species that reduces N S , we explored the impact of I 3 − on N S using AIMD simulations.The interfacial concentration profiles at the N 2 gas-D 2 O solution of NaI 3 interfaces with various NaI 3 concentrations are plotted in Figure 3a−c, while the variation of N S is plotted in Figure 3d.The interfacial concentration profiles reveal an extremely high surface propensity of I 3 − at the N 2 gas−water interface.Such a high propensity reduces N S , as is evident from Figure 3d; when the surface density of I 3 − is 2.0 nm −2 , N S decreases by ∼60%.This indicates that the contact area of the N 2 gas−water interface can be reduced due to the presence of I 3 − .The higher surface propensity of I 3 − than I − stems from a much lower negative charge density of I 3 − than that of I − . 61he interfacial concentration profiles obtained from the AIMD simulation suggest that [NaI 3 ] is high at the interfaces but is extremely low in the bulk.This implies that any I 3 − present in solution will appear at the surface, and trace amounts of − concentration to be ∼47 and ∼54 nM at [NaI] = 5 and 6 M, respectively. 62The AIMD simulation indicates that such extremely low concentrations of I 3 − in the bulk are sufficient to modify the water surface since, in the simulations, the density of I 3 − is near-zero in the z < −5 Å region (bulk region).Furthermore, by computing the energy

Environmental Science & Technology
difference between the partially hydrated and fully hydrated states of I 3 − , we infer the population difference between the interface and bulk to be different by a factor of 3 × 10 8 (see Supporting Information Section S8) to predict the surface density of I 3 − at the corresponding [NaI] in bulk (Figure 3d).As is clear from the quantitative comparison in Figure 3d, the AIMD simulation of NaI 3 captures the trend from the experiment.
To confirm that the I 3 − species critically affects the free O−D peak amplitude, we compared Imχ eff,ssp (2) spectra from the surface of a 1.
In contrast, Na 2 S 2 O 3 is known to serve as a strong reductant.Thus, the addition of Na The Imχ yyz (2) spectra obtained from the measured Imχ eff,ssp spectra are displayed in Figure 4a−c.The Gaussian fit to the spectra (Supporting Information) provides the amplitude of the free O−D peak.The amplitude for the NaI solution is  (0.98 ± 0.03) × 10 −21 m 2 /V, while it decreases to (0.75 ± 0.04) × 10 −21 m 2 /V upon the addition of I 2 .By reducing the I 2 -containing sample using Na 2 S 2 O 3 , the amplitude was recovered to (0.99 ± 0.07) × 10 −21 m 2 /V, i.e., equal to the amplitude of pure NaI solution within experimental uncertainty.Note that this recovery took more than 1 h, presumably because of the low surface activity of Na 2 S 2 O 3 .This observation is consistent with the scenario described above, manifesting that the free O−D reduction arises from I 3

−
and not from I − (Figure 4d).Our data witness the extreme surface propensity of I 3 − at the N 2 gas−water interface and the absence of I − at the topmost N 2 gas−water interface.This appearance of I 3 − occurs via the salting-out of large hydrophobic ions (I 3 − ) by small hydrophilic ions (I − ). 63Such insights into the ion composition and distribution in aerosols are crucial because they determine the chemistry occurring inside and on the surface of aerosols.This chemistry also depends on the aerosol size, which governs the ratio of surface area to the number of I 3 − , a molecular reservoir of I 2. Since I 3 − is localized in the outermost aerosol layer, it can emit I 2 into the vapor phase via the backward reaction (Figure 4d).This chemical pathway may contribute to I 2 emission into the atmosphere, in addition to the well-known I 2 emission from I − and HOI. 3,64inally, we note other experimental techniques that can shed light on this challenge.Second harmonic generation spectroscopy 65 and electronic SFG 66−68 may be able to probe I 3 − if the visible beam wavelength is set to match the optical transition of I 3 − and the Raman cross section is sufficiently large.Probing different iodine species across a wide wavelength region may clarify the iodine chemistry at the interface.Liquid jet X-ray photoelectron spectroscopy is another powerful tool to probe surface composition, but this technique may give different results compared to this SFG study, as in the liquid jet, the surface is generated quasi-instantaneously, and we expect the generation of I 3 − to be slower than the time scale of the surface generation in these types of experiments. 35,38ENVIRONMENTAL IMPLICATIONS Here, by combining PD-HD-SFG spectroscopy with AIMD simulation, we confirmed that the iodide ion does not significantly reduce the contact area of the N 2 gas−water interface; iodide ions are instead located below the topmost water layer.When the iodide ion concentration increases, the triiodide ions are more present at the vapor−water interface, resulting in a ∼10 8 times higher concentration at the surface than in bulk.−72 This suggests that the vapor−water interface provides a unique sink and reaction environment for triiodide ions.Our result demonstrates that the reaction mechanism and stability of the polyatomic iodine differ significantly between the bulk and at the interface.Together with the salting-out of triiodide by other ion species in the sea, our current study indicates that triiodide can be present on the surfaces even without sunlight (at night).
Experimental and Simulation Methods.Sample Preparation.Sodium iodide (>99.5%) was purchased from Alfa Aesar.Sodium iodate (>99.5%),iodine (>99.99%), and D 2 O (>99.9%) were obtained from Sigma-Aldrich.Sodium thiosulfate (≥99%, anhydrous) was obtained from Carl Roth.These materials were used as received.We used H 2 O obtained from a Milli-Q machine (a resistance of 18.2 MΩ cm).To avoid oxidation of the iodide ion as much as possible, we dissolved sodium iodide salt into D 2 O (H 2 O) under an N 2 atmosphere and in a dark room just before SFG (UV−visible absorption) experiments.We poured the resultant D 2 O solution of NaI into a PTFE dish with a diameter of 6 cm and measured the SFG spectra from the samples.We emphasize that we avoided the laser irradiation for the samples before SFG spectra measurements.For generating I 3 − ions, we dissolved I 3 − (50 mg) into NaI solution (1.5 M, 20 mL).The concentration of the generated I 3 − ion was calculated to be ∼10 mM based on the equilibrium constant of ∼700 M −1 for the triiodide formation reaction of I − + I 2 ⇄ I 3 − . 73By diluting this solution with a 1.5 M NaI solution, we obtained a 1.5 M NaI solution with 1 μM I 2 and used it in Figure 4b.For NaI solution with 1 μM I 2 and 100 mM Na 2 S 2 O 3 , we first dissolved I 2 (50 mg) into Na 2 S 2 O 3 solution (100 mM, 20 mL) to exclude the possibility that a trace amount of I 2 remained in the solution.After that, we diluted this solution with a NaI and Na 2 S 2 O 3 mixture solution to have 1 μM I 2 and used it in Figure 4c.Note that the sequential addition of Na 2 S 2 O 3 into the NaI solution with 1 μM I 2 gave a similar trend in the recovery of the free O−D contribution.
HD-SFG Measurement.The details of our PD-HD-SFG setup are presented in ref 74.Briefly, we focused the infrared (IR) and visible beams collinearly onto a y-cut quartz to generate a local oscillator (LO) signal.A 5 mm-thick SrTiO 3 plate was inserted into the beam path to generate the delay for the LO beam relative to the other beams.These beams were refocused onto the N 2 gas-D 2 O solution at the NaI interface.The angles of incidence were set to 45°with respect to the surface normal.We used ssp and ppp polarization combinations, where ssp (ppp) denotes s-(p-)polarized SFG, s-(p-)polarized visible, and p-(p-)polarized IR beams.The measurements were performed under a nitrogen atmosphere with a humidity of less than 0.1%.After 15 min of equilibration of the samples under a nitrogen atmosphere, the HD-SFG measurements were carried out.The exposure time of the measurement was set to three min, and the data obtained was averaged over 15 min.Note that, over the course of the measurements, we did not observe changes in the spectral feature within this time scale.Freshly prepared NaI solutions were used to prevent oxidation due to oxygen in the air.Further details of the setup and HD-SFG measurements are present in the Supporting Information.
AIMDs Simulation.AIMD simulations for the systems of pure D 2 O, NaI in D 2 O, and NaI 3 in D 2 O have been conducted with the CP2K code. 75The system of pure D −1 band originate from free O−D groups, the antisymmetric O−D stretch mode of the D 2 O molecules donating two hydrogen bonds, and hydrogen-bonded O−D groups, respectively. 47−49 Upon the addition of NaI, the Imχ eff,ssp (2) amplitude decreases above 2560 cm −1 , while it increases below 2560 cm −1 , consistent with refs 29 and 50.The Imχ eff,ppp (2) spectra commonly show a 2730 cm −1 positive peak, a 2650 cm −1 negative band, and a broad 2550 cm −1 positive band.Upon increasing [NaI], the 2730 cm −1 contribution does not change significantly, unlike in the Imχ eff,ssp

Figure 1 .
Figure 1.PD-HD-SFG measurement at the N 2 gas-D 2 O interface.(a) Imχ eff,ssp (2) and (b) Imχ eff,ppp we obtained N S as a function of [NaI], as displayed in Figure 2a.This figure indicates that N S decreases drastically with increasing [NaI], particularly at elevated [NaI].The decrease in N S results from the reduction of the contact area of the N 2 gas−water interface due to increasing the number of ions at the interface.Note that since the change in the peak area A yyz arises not only from the change in N S upon the variation of [NaI] but also from the change of the orientation of the free O−D group, as is discussed in Supporting Information Section S3.4, the careful separation of the contributions from the variation of ⟨cosθ⟩ and ⟨cos 3 θ⟩ and from N S through the PD-HD-SFG was needed.
2 O interface ((D 2 O) 260 ) and vapor-D 2 O solutions of NaI, (D 2 O) 236 (NaI) 12 , and (D 2 O) 216 (NaI) 24 , and we computed the variations of N S of the free O−D group from the AIMD trajectories.The simulation details can be found in the Supporting Information.The obtained N S vs [NaI] data is plotted in Figure 2a, where the values of [NaI] were obtained from the interfacial concentration profiles (Figure 2b−d) in the z < −7 Å region (bulk region).Note that the origin point of the z-axis is the position of the Gibbs dividing surface.The discrepancy

I 3 −
are sufficient to modify the surface density of the free O−D groups.To examine whether any I 3 − is present in our samples, we measured bulk ultraviolet−visible (UV−vis) spectra of the NaI−H 2 O solutions with various [NaI].The data are displayed in Figure 3e.A 350 nm feature, a fingerprint of the I 3 − species, is not detectable for [NaI] < 4 M.For [NaI] > 4 M, a tiny 350 nm feature is apparent.From the UV−vis data, we infer an upper limit of I 3

Figure 2 .
Figure 2. Surface population of the free O−D chromophores as a function of [NaI] in bulk.(a) Normalized surface density of the free O−D chromophores N S as a function of [NaI].N S values are normalized to the value for the bare surface.The shaded region around the experimental data represents the error bars obtained from the fits and subsequent orientational analysis (Supporting Information); the error bars on the simulation data represent the 95% confidence intervals obtained from multiple data sets.The red dotted line shows the weak variation of the free O−D population expected from AIMD simulations.(b−d) Computed interfacial concentration profiles at the vapor−water interface in the absence (b) and presence of NaI (c,d).The water concentration profiles are rescaled by a factor of 0.1.Zero depth is the position of the Gibbs dividing surface for water.

5 M
NaI solution, a 1.5 M NaI + 1 μM I 2 solution, and a 1.5 M NaI + 1 μM I 2 + 100 mM Na 2 S 2 O 3 solution.If the reduction of the free O−D peak arises from I 3 − , the addition of I 2 to the NaI solution should lead to the reduction of the free O−D peak via the generation of I 3

3 − 3 −
2 S 2 O 3 to the solution containing I is reduced, one can expect that the free O−D peak for the 1.5 M NaI + 1 μM I 2 + 100 mM Na 2 S 2 O 3 solution is comparable to the free O−D peak for the 1.5 M NaI solution.

Figure 3 .
Figure 3. Variation of the surface density of the free O−D chromophores in the presence of I 3 − .(a−c) Interfacial concentration profiles for NaI 3 solutions with various concentrations.The water concentration profile was rescaled by a factor of 0.1 for clarity.(d) Comparison of the variations of free O−D populations between the simulation and experiment.The experimental data are identical to the data in Figure 2a.The surface density of I 3 − for the experimental data was estimated from the UV−vis spectra and AIMD simulation (see Supporting Information Sections S7 and S8).The red dotted line serves to guide the eyes.(e) UV−vis spectra for NaI samples with varying bulk concentrations of NaI, together with the absorption spectrum for the 100 μM I 3 − ion (dotted line).The absorption spectrum of I 3 − was scaled by a factor of 0.01.The black solid lines represent the baseline computed based on the fifth-order polynomial function to clarify the appearance of the 350 nm feature.

Figure 4 .
Figure 4. Effect of chemically generated I 3 − on the free O−D amplitude.(a−c), Imχ yyz (2) spectra at the N 2 gas-D 2 O solutions interfaces: 1.5 M NaI D 2 O solution (a), 1.5 M NaI and 1 μM I 2 D 2 O solution (b), and 1.5 M NaI, 1 μM I 2, and 100 mM Na 2 S 2 O 3 D 2 O solution (c).The solid lines represent the fits, and the dotted lines indicate the peak contributions obtained from the fits.The free O−D peak feature is highlighted by filling in the dotted lines.The arrows show the amplitude of the free O−D peak.(d) Schematic representation of the ion organization at the vapor-aqueous solution of the NaI interface.I − does not perturb the topmost water structure and thus is present below the topmost water layer.On the other hand, I 3 − is present in the topmost water layer due to its high surface propensity.I 3 − can serve as a reservoir/emitter of I 2 at the interface via the reaction I − + I 2 ⇄ I 3 − . 62The error bars indicate the 95% confidence intervals of the multiple measurements.
2 O consisted of 260 D 2 O molecules, while the systems of NaI in D 2 O consisted of 216 D 2 O and 24 NaI [(D 2 O) 216 (NaI) 24 ] and 236 D 2 O and 12 NaI [(D 2 O) 236 (NaI) 12 ].These molecules were contained in the 14.4 × 14.4 × 70.0 Å cell.The systems of NaI 3 in D 2 O consisted of 114 D 2 O and 8 NaI 3 [(D 2 O) 114 (NaI 3 ) 8 ], 114 D 2 O and 4 NaI 3 [(D 2 O) 114 (NaI 3 ) 4 ], and 114 D 2 O and 2 NaI 3 [(D 2 O) 114 (NaI 3 ) 2 ].These molecules were contained in the 14.4 × 14.4 × 50.0 Å cell.For the pure D 2 O and NaI in D 2 O systems, we prepared the 5 random structures for the pure Environmental Science & Technology D 2 O system and the 10 random structures for the NaI in D 2 O and NaI 3 in D 2 O systems using the Packmol code. 76After 1 ns of equilibration in the force field MD simulation, we performed an AIMD simulation for these samples.Further details of the procedure can be found in the Supporting Information.■ ASSOCIATED CONTENT * sı Supporting Information The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c08243.Details of AIMD simulation; definition of the free O−D group for NaI and NaI 3 solutions; protocol for interfacial and bulk concentrations of I − and I 3 − ions; experimental details for HD-SFG setup and UV−visible absorption measurement; amplitude calibration of the HD-SFG spectra and Fresnel factor correction; procedures for the fits of HD-SFG spectra and for calculating the orientational and density analysis of the free O−D group; surface pressure data of NaCl and NaI solutions; comparison of NaBr and NaI samples; comparison of the neat D 2 O and NaIO 3 samples; energy difference of the I 3 − molecule at the water−air interface and in the bulk; and photosensitivity of the NaI sample under SFG measurements (PDF)■ AUTHORINFORMATION Mischa Bonn − Max Planck Institute for Polymer Research, 55128 Mainz, Germany; orcid.org/0000-0001-6851-8453;Email: bonn@mpip-mainz.mpg.deYuki Nagata − Max Planck Institute for Polymer Research, 55128 Mainz, Germany; orcid.org/0000-0001-9727-6641;Email: nagata@mpip-mainz.mpg.de