VSFG spectroscopy is a laboratory based analysis method, hence 50 mL SML samples from the mesocosm and open-sea studies were stored in polyethylen bottles at -18°C and analyzed within a few weeks. Preliminary experiments showed that the measured SFG spectra of samples frozen for many weeks did not differ from those of fresh, unfrozen samples. The defrosted samples were thoroughly mixed and aliquots of 10 mL were quickly transferred to the teflon dish. The samples were allowed to rest for about 10 min before starting an SFG spectral scan by tuning the IR laser frequency in steps of 5 cm-1 over the C-H stretch vibration wavenumber range of 2800–3000 cm-1. Selected samples from the Helgoland cruise were also analyzed in the wavenumber range 1650–1800 cm-1, which is typically assigned to the C=O stretch vibrations of carbonyl groups. The determination of the surface coverage sc from the average VSFG spectrum of the 3–5 aliquotes for each sample will be described in more detail in Sect. . Within the limits of the model's validity and derived from error propagation using the 1σ repeatability of the individual spectra (6 %), the analytical precision of sc is assessed to be 10 % (2σ). Surfactant activity was measured with an analytical precision of <10 % by AC voltammetry (METROHM 797 VA Computrace) with a hanging drop mercury electrode according to a protocol developed by and modified with an internal calibration technique . All SML samples were collected in polyethylene bottles (50mL) and stored at 4 °C. 10mL of the unfiltered samples were analyzed in triplicate using a standard addition technique. Calibration via standard addition was performed by the addition of TX-100 stock solution, where the surfactant activity is given as µg TX-100 equivalent L-1(µgTeqL-1). Sodium chloride solution (NaCl) with a concentration of 35gL-1 was used as blanks.

A large-scale algal bloom mesocosm experiment has been conducted in the Sea Surface Facility (SURF) at the Institute for Chemistry and Biology of the Marine Environment in Wilhelmshaven, Germany, from 18 May to 16 June in 2023. The experiment was part of the BASS (biogeochemical processes and air-sea exchange in the sea-surface microlayer) project targeting the role of the SML in biogeochemical cycles and air-sea interaction. The details and related studies of the mesocosm campaign are described elsewhere . The induced phytoplankton bloom with a distinct succession of pre-bloom, bloom, and post-bloom phase led to high surfactant concentration with distinct slick formation and surfactant activities up to 1868µgTeqL-1 in the SML, hence exceeding that observed in natural environments. In brief, the experiment was conducted as follows. The 8.5m long, 2m wide, and 1m deep concrete pool was filled with 17000L of pretreated salt water from the Jade Bay (53°28′42′′ N, 8°12′15′′ E). The pool was equipped with eight flow pumps (AQUALIGHT, ATK-4 Wavemaker, 18000Lh-1) at the bottom of the pool to ensure homogeneity of the water column and to prevent particle settling and biofilm formation on the pool walls. All pumps were operated at low power settings to mimic a slow natural mixing in the bulk water and its surface without turbulent mixing. The facility was exposed to ambient light and was protected from raining events by a retractable 4mm polycarbonate roof (highly transmissible for sunlight). Prior to the start of the time series, the water surface was cleaned of organic matter by extensive glass plate sampling over 9h. Throughout the experiment, nutrients were added to the tank on three different days (26, 31 May and 1 June 2023) to achieve final concentrations of 0, 10, and 5µmolL-1 for nitrogen, 1.2, 0.6, and 0.3µmolL-1 for phosphorus and 19.8, 10, and 0µmolL-1 for silicate, respectively. The samples shown in this study were collected daily, alternating between morning (around 6 a.m.) and afternoon (around 4 p.m.) based on the sunrise timing. The SML samples were taken using the glass plate method according to a protocol outlined in .

The samples were taking during the BASS Helgoland field campaign (RV Heincke, cruise no. HE644) taking place from 9 June to 1 August in 2024. All SML samples were collected off-shore of the island Helgoland (North Sea, around 54°15′ N, 8°00′ E) by remote-controlled catamarans using a rotating sampler, either the autonomous surface vehicle HALOBATES or the radio-controlled vehicle GLAUCUS. Additionally, underlying water samples were collected at 1m depth. The field campaign target was to spot patches of SML slick and perform sampling inside and outside the slick patch to elucidate differences in the SML, for this study in particular the surfactant state. Ideally, a sampling series of a given sampling event consisted of 9 distinct samples (3 bulk+3 SML slick+3 SML no slick). The SML samples are qualitatively categorized into SML (slick), SML (non-slick), whether or not a reduction of the surface tension of the water – becoming visible through a pronounced damping of the wave field – was observed during sampling.

Triton X-100 is widely used in the marine science community as a reference compound to measure surfactant activity by means of AC voltammetry and to study the effect of a soluble surfactant on air-sea gas exchange . To the best of our knowledge, there are no VSFG studies available in the literature addressing the surfactant behavior of TX-100. Figure a displays VSFG spectra of TX-100 for increasing bulk concentrations. Already at the lowest concentration of 50µgL-1, clear spectral signatures are visible, indicating that the air-water interface is already covered by a significant fraction by TX-100 molecules. This is consistent with measurements that have reported a pronounced inhibition of air-sea gas transfer already at such low concentrations . As outlined above, the observed vibrational bands at 2850, 2880, and 2950cm-1 can be assigned to d+,r+, and rFR. Compared to common VSFG spectra of fatty acids with longer alkyl chains such as DPPC, stearic acid or oleic acid , the observed vibrational bands are broader, probably due to variable line shifts resulting from both mesomeric and inductive effects acting on the different CH2 and CH3 groups in the molecule (see Fig. ). Here, the presence of the ether group in the polyether tail is expected to induce a blue shift of CH2 group directly bound to the ether group, and the positive inductive effects of the alkyl groups in the hydrophobic part of the molecule should lead to a red shift of C-H signatures of the octyl group. Quite unusual for VSFG spectra measured in the ssp polarization combination is the distinct band at 2915 cm-1, which can be assigned to the antisymmetric stretch vibration of the methylene group (CH2, d- mode). d- is often not detected for molecules with longer alkyl chains because (i) the pseudo-inversion symmetry of apposing facing methylene groups in highly ordered alkyl chains in all-trans conformation cancels out the corresponding SFG intensities and (ii) the net contribution of the this vibration to the transition dipole moment perpendicular to the interface remains small. In principle, d- intensity in TX-100 can arise from both the CH2 groups in the polyether chain and the hydrophobic octyl part of the molecule. At low surface concentrations, the polyether chains lie flat on the surface and thus contribute to the signal. Such a structure is supported by molecular dynamics simulations of (see Fig. S9 in their Supplement). However, as the surface concentration increases, the polyether groups are expected to get displaced from the interface and get integrated into the water network. This is expected to reduce the overall structural order, consequently diminishing the intensity contributions to the VSFG signal. Note, however, that the d- peak remains prominent in all spectra, suggesting that a CH2 group maintains a preferred orientation even when the surface concentration is high. Tentatively, we attribute this strong signal component to the single CH2 in the hydrophobic part of the molecule as it may result from the packing pattern of the bulky alkylbenzene group on the surface. Future polarisation dependent measurements are needed to further elucidate such interesting structural effects .

In order to link the measured SFG intensity to surface concentration and surface coverage, Fig. b directly compares the measured surface pressure at T=293 K (filled green circles) with the VSFG spectral area aSFG-CH (black squares) as function of the volumetric TX-100 concentration in the bulk. The open green symbols correspond to surface pressure data taken from (interpolated to 293 K) and (room temperature) for comparison. The three data sets are largely consistent but do not match up entirely. This could be attributed to minor variations in the compositions of the TX-100 batches used (comprising molecules with differing polyether chain lengths), alongside residual impacts from temperature and pH variations. However, both the surface-pressure data as well as VSFG curves reveal a Langmuir-type behavior, demonstrating the adsorption equilibrium of TX-100 between the volumetric water phase and the interface. The Langmuir isotherm can be written as 4ΓΓmax≈cscmaxs=θθmax=c⋅KL1+c⋅KL≈scscmax with KL the Langmuir adsorption constant quantifying the affinity of a molecule for adsorption at an interface. Γ denotes the surface excess (i.e. the additional amount of a component per unit area that accumulates at an interface compared to its bulk concentration), cs the surface concentration (in units of molecules/area), θ the number of available adsorption sites per unit area, and sc the surface coverage as defined in this work. The corresponding maximum values in the denominator are the limiting values for high surfactant concentration c in the bulk. As we shall demonstrate in the following sections, scmax≈100% is applicable to an interface completely covered with natural surfactants.

In order to convert the volumetric surfactant concentration c to a surface concentration cs, we take advantage of the Gibbs adsorption equation for an ideal dilute solution, providing the link between surface tension γ and surface concentration in terms of surface excess. 5cs≈Γ=-1RTdγdln⁡cT Here, R is the gas constant and T the absolute temperature. The combination of the Langmuir and Gibbs equations (Eqs.  and ) leads to the Szyszkowski equation , which enables the calculation of surface pressure Π=γ0-γ, with γ0 the surface tension of a surfactant-free water interface. 6Π=RTΓmaxln⁡1+KLc The Langmuir constant KL and the maximum surface excess Γmax are frequently derived from fitting experimental surface pressure isotherms. In case of the rather strongly adsorbing TX-100 (KL=1.5×103m3mol-1,Γmax=2.9×10-6molm-2, ), the fraction of bulk molecules within a monolayer thickness of δ=2nm is small compared to the surface excess. Hence, surface excess equals surface concentration (cs=Γ+cδ≈Γ) in this case.

Building on this thermodynamic framework, Fig. c shows the same isotherms as in Fig. b, but plotted as function of the surface concentration. While the surface pressure isotherm exhibits a continuous rise in surface pressure, a behavior characteristic of soluble surfactants, the associated VSFG signal intensity indicates a shoulder or a plateau-like trend at surface concentrations of approximately cs≈1.2moleculenm-2 (corresponding to a free molecular area of 83Å2molecule-1). Such a significant change in the gradient of the SFG intensity curve often indicates a 2D phase transformation of the structure or at least a significant re-structuring of the molecules in the organic monolayer. Consequently, based on the interpretation of the VSFG spectrum outline above, the signal plateau corresponds to the surface concentration level where the interactions among polyethylene chains lead to their detachment from the interface. This surface concentration can be identified as a surface almost completely covered with organic molecules. The dashed gray line in Fig. c corresponds to the SFG signal intensity that we set equivalent to a surface coverage of sc=100%, actually based on DPPC spectra discussed in the next section. Note that this signal level is consistent with the end of the plateau region of TX-100. Finally, at surface concentrations corresponding to free molecular areas below 67Å2molecule-1 a steep increase of SFG intensity is observed. This points at a densely packed 2D phase state with high molecular order, again consistent with the expected maximum possible surface concentration of a stable monolayer of TX-100 with a collapse point at a molecular area of about 66Å2molecule-1 .

DPPC is a well-studied insoluble surfactant with a typical 2D phase behavior of a phospholipid or fatty acid . The 2D phase behavior is clearly visible in VSFG spectra as well . Figure a shows a series of VSFG spectra with increasing surface pressure. The symmetric r+ and d+ vibrations of the CH2 and CH3 groups yield well resolved peaks at 2850 and 2875 cm-1, going along with the corresponding overlapping Fermi resonances around 2955 cm-1. With increasing surface pressure, the intensity of the CH3(r+) vibration increases, while the CH2(d+) vibration slightly decreases. The resulting decrease of the CH2/CH3 ratio is associated with the increasingly uniform all-trans alignment of the alkyl chains . In Fig. b, the measured Langmuir isotherm of DPPC is shown together with the integrated SFG signal intensity aSFG-CH (black symbols and interpolated curve). Both the surface-pressure isotherm and the SFG signal trend can be interpreted in the context of 2D phase transitions. For example, at surface concentrations below the lift-off point, the insoluble surfactant DPPC has an almost negligible impact on surface tension (Π<0.2mNm-1 from 1.0–1.25moleculesnm-2). This is in striking contrast to soluble surfactants showing a gradual decrease of the surface tension as outlined for TX-100 in the preceding section. The lift-off point at 80Å2molecule-1 separates the 2D gas phase from the liquid expanded (LE) phase. It corresponds a situation where lateral interactions between neighboring molecules become noticeable. This situation can phenomenologically be identified with a surface being loosely covered with surfactants. Towards higher surface concentration, the plateau of the surface pressure between 66Å2molecule-1 and 50Å2molecule-1 can be attributed to a phase transition from the liquid expanded (LE) phase into liquid condensed (LC) phase. Here, the molecules start to avoid unfavorable intermolecular interactions by the reorientation of alkyl chains. This reduces repulsive interactions from steric hindrance and enhances favorable interactions such as van der Waals forces between densely packed alkyl chains. Phenomenologically, it is reasonable to identify this situation with a surface being completely covered with surfactants, hence sc=100%. Note that the phase transition to the liquid condensed 2D phase is going along with a plateau-like SFG signal intensity (Fig. b), again highlighting that the phase transition corresponds to a monolayer compression going along with significant structural re-organization of the surfactant molecules. Consequently, the corresponding integral signal intensity at a surface concentration of 1.45moleculesnm-2(69Å2molecule-1),aref, serves as a suitable reference point to scale the surface coverage parameter sc=aSFG-CH/aref to 100 %.

Note that in previous work , we have chosen a higher surface concentration of DPPC, 40Å2molecule-1, as the reference point. This alternative reference point with an about 2.6 times higher aSFG-CH value corresponds to the leveling off of the SFG intensity. It is close to the collapse point of the monolayer, where DPPC is in its highly compressed solid 2D phase state with straight and perfectly aligned alkyl chains. Considering (i) that such a well-ordered monolayer is unlikely to be representative of a natural SML nanolayer and (ii) that a pronounced SFG signal enhancement is induced by the preferred orientation of the transition moment of the terminal methyl groups of the lipid chain, we consider the plateau signal at about 69Å2molecule-1 as a more practical choice for naturally existing SML interfaces. The corresponding signal level is indicated as a horizontal dashed line both in Fig.  for DPPC and in Fig.  for TX-100. This choice has advantages and limitations. Firstly, this reference point is relatively easy to reproduce experimentally and, secondly, it represents a not perfectly structured monolayer. This is realistic for natural systems, which typically consist of complex mixtures of more or less soluble lipids, carbohydrates, proteins, and other organic compounds. The diversity in molecular substance classes, molecular size and structure is expected to induce a certain degree of disorder. Thirdly, it turns out that the corresponding aref value for DPPC is fully consistent with the observed SFG plateau level of TX-100 that has been identified with a completely covered surface as well. This seems all very consistent in itself, nevertheless is important to emphasize that the selection of the reference point remains somewhat arbitrary and is guided by such practical considerations. Therefore, the operational definition of the surface coverage parameter sc should not be taken as a universal thermodynamic threshold applicable to all surfactants or interfacial systems. Rather the objective of this approach is to establish a surface coverage parameter that is easily comprehensible and directly applicable for the characterization of SML samples through SFG spectroscopy, bypassing intricate spectroscopic details.

In the SURF mesocosm experiment, a nutrient-induced phytoplankton bloom development in North Sea water has been investigated to gain a comprehensive understanding on the dynamic coupling of biogeochemical processes between the SML and the underlying water (ULW). We provided data on surface coverage and surface activity, which have been discussed in a complementary overview paper in the context of biological processes that fed into the surfactant pool . Here, we focus on the direct comparison of surface coverage and surface activity data during the three main phases of the experiments (18–26 May: pre-bloom, 27 May–4 June: bloom, 5–16 June: post-bloom), which are well reflected by the abundance of chlorophyll a (chl a) shown as a green shaded curve in Fig.  . The black square symbols correspond to the daily measured SML surface coverage data as derived from the VSFG spectrum using Eqs. () and () and the red and blue square symbols to the SML and ULW surfactant activity values in TX-100 equivalents. The solid curves correspond to a spline interpolation to guide the eye. All data are listed in Table S1 in the Supplement together with uncertainty limits representing the analytical precision of the replicated measurements. Data are also available in PANGAEA , and have been discussed previously in the context of biological drivers for surfactant formation in .

Initially, the surface coverage was very low (<10%), as expected after the extensive cleaning of the surface prior to the start of the experiment. From 20–26 May, still in the pre-bloom phase, a slight increase in surface coverage can be observed, averaging 31 %. After the first nutrient addition on 26 May, a strong increase was observed, with surface coverage reaching values above 80 % within three days. For the remainder of the experiment, surface coverage remained high. The slight increase towards the end of the experiment is within the uncertainty limits, suggesting that the nanolayer quickly approached a saturation level, latest at the end of the bloom phase. The observed surfactant coverage trend is in surprisingly stark contrast with the surfactant activity data. Here, the ULW data show almost constant surfactant activity throughout the experiment, with mean values for the pre-bloom, bloom, and post-bloom phases of 174, 184, and 205µgTeqL-1, respectively. In contrast, the SML surfactant activity increases substantially in response to the algal bloom. From the pre-bloom to early bloom phase, surfactant activity increases from roughly 200–400µgTeqL-1, paralleling the rise in surface coverage observed up to 31 May. In the late bloom and post-bloom phases, SML surfactant activity reaches its maximum, peaking at about 2000µgTeqL-1, with a few days delay compared to the chl a data. While surfactant activity decreases after this peak, it remains elevated, with most post-bloom measurements above 400µgTeqL-1.

The surfactant data clearly reveal a bloom-induced enrichment of surfactants in the SML, either by effective ULW to SML transport or by enhanced formation in the SML slick acting as a biofilm-like habitat . Indeed, reported that the SML was enriched in bacterial cells during the bloom phase, whereas bacterial abundance in the ULW remained relatively low, followed by higher abundance of bacterial cells during the post-bloom phase in both SML and ULW. Biosurfactants are formed, in particular, by microbial processing of organic matter , where the delay of the surfactant peak compared to chl a aligns well with the utilization of phytoplankton derived OM. Mismatches of surfactant abundance and primary productivity have been reported previously and are in line with surfactants originating from phytoplankton exudates or cellular lysis . The decline in surfactant activity observed towards the end of the experiment is probably due to ongoing microbial activity or photochemical breakdown of surfactants. Light exposure remained consistently high in the later phase of the experiment, and the carbon utilization assay indicated that bacterial activities were increasingly targeting carbohydrates as substrates for respiration , suggesting a modification in the molecular composition of the available OM pool.

Overall, the direct comparison of the data clearly shows that surfactant activity and surfactant coverage serve as complementary sum parameters. Whereas surface activity is a measure of surfactant availability in the bulk SML, the surface coverage parameter sc provides direct information about the equilibrium of surfactants between the bulk SML and the nanolayer. For the OM pool present in this study, it is evident that already at surfactant activity values of 400µgTeqL-1, the surface is almost completely covered with surface-active substances, at least in the non-disturbed equilibrium that was established under our laboratory conditions with resting water samples. Even the slight increase in surface activity at the beginning of the bloom phase around 27 May was sufficient to saturate the surface with surfactants. Until the end of the experiment, the surfactant activity did not fall significantly below this value again and the SML essentially represented a slick layer. Certainly, these persistent conditions can be ascribed to the fact that the experiment was carried out in the confined environment of a mesocosm pool. While the potential bias resulting from the accumulation of insoluble surfactants on the surface was mitigated by their selective removal during routine plate sampling, under open ocean conditions, wave motion and the extensive surface area would typically inhibit the establishment of persistent slicks.

However, the surfactant activities reported here are not out of the range. Even for open ocean conditions, typical values range from 100 to 600µgTeqL-1 in SML samples , and a general threshold value of 1000 µgTeqL-1 has been suggested for visible/strong slick events . The latter is fully compatible with our threshold value of about 400µgTeqL-1 indicating a completely (>80 %) surfactant covered surface. Data from a wind–wave tunnel experiment with natural sea water from have shown that surface coverages above this threshold correspond to conditions associated with strong reduction in the mean square slope of surface waves, whereas surface coverages below 50 % had an overall small effect. When applied to our dataset, this could indicate a pronounced shift between surfactant conditions which would induce low and high wave damping effects. While pre-bloom and ULW concentrations of 200µgTeqL-1 may fall short of inducing wave damping, concentrations rapidly surpassing 400µgTeqL-1 in the SML during the bloom phase could be very effective. This consideration suggests that even moderate increases in SML surfactant activity may disproportionately enhance surface coverage and wave damping effects , reinforcing the importance of including surfactant microlayer dynamics in future air–sea gas exchange parameterizations.

Our research focus of the Helgoland field campaign was to compare the magnitude of surface coverage values obtained from the mesocosm study with those from natural open ocean samples on the one hand, and to assess surface coverage differences under slick and non-slick conditions relative to the ULW on the other hand.

Figure a shows the surface coverage data for the three sample categories using a box-and-whisker plot. Corresponding single surface coverage and surfactant activity values are listed in Table S2 in the Supplement. The data resemble days where patches of slick events have been observed and samples could be classified as slick, non-slick and ULW according to visible or not visible wave attenuation. Although the overall linear correlation between surfactant activity and surface coverage was weak (R2=0.31, p=0.03), a clear trend in median surface coverages with 72 % (SML slick) >47 % (SML non-slick) >23 % ULW is evident. We attribute the wide 25 %–75 % interquartile ranges (IQR) at least partly to uncertainties arising from the somewhat subjective classification scheme for slick and non-slick conditions, but also to substantial natural variability, likely driven by fluctuations in wind, weather, oceanic fronts, and biological activity. The significantly lower ULW values are in line with expectation that surfactants are accumulated in the SML. Moreover, the comparison of surface coverage for ULW and SML (non-slick and slick samples) at comparable surfactant activities revealed a factor of about 2.5 higher surface coverages for the SML samples, suggesting that the surfactant pool in the SML may differ from that in the ULW. Further evidence of a changing molecular composition of the different samples is provided by VSFG spectral signatures. Whereas uniform spectra have been measured in the C-H wavenumber range 2800–3000cm-1, a notable spectral feature was observed in the C=O stretch vibration range, which is typical for carbonyl groups (1650–1800cm-1). Figure b compares averaged VSFG spectra from the ULW and SML (slick and non-slick) samples. The vibrational bands at 1720 and 1740cm-1 can be assigned to aldehydes and ketones or ester and acid functional groups, respectively. The intensity of the spectra show the same trend as the surface coverage, SML (slick)>SML (non-slick)>ULW, but the signal enhancement in the slick samples is much more pronounced than one would expect based on the surface coverage ratios alone. In principle, this could indicate enhanced photochemical or oxidative processing of SML surfactants in slicks, which has been reported in previous studies . However, the sensitivity of SFG with respect to molecular orientation effects (Eq. ) may have lead to higher signal intensities in the dense organic monolayer of the slick samples as well.

Overall, the observed surface coverage values are comparable to those obtained in the mesocosm experiment and the highest measured values are fully compatible with our definition of 100 % coverage, further supporting the validity of the chosen surface coverage reference point. Interestingly, even the non-slick samples exhibited relatively high surface coverages on average. Given that the distinction between slick and non-slick was made purely on visual grounds, this suggests that a significant, visible wave attenuation only occurs for essentially completely surfactant covered surfaces. The high values of the non-slick samples also raise the question of whether such high values could be normal under open sea conditions. To come closer to answering this question, we first need to establish a link between surface coverage and the easier accessible surfactant activity measure.