Introduction

Nanotechnology is an emerging field that has a significant role in all sectors of life, including agriculture [1]. A wide diversity of impacts by nanomaterials on plants and agriculture has been well documented, which proves that positive and negative effects depend on their different properties [2,3,4,5,6,7,8].

The efficiency of bulk iron oxides in providing iron to plants has appreciably grown after their conversion into nanoparticles [3, 8]. These nanoparticles are acquiring an important place in the development of agriculture and other industries, due to their unique properties, such as quantum confinement, high surface area to volume ratio and catalytic properties [9, 10]. In fact, iron oxide nanoparticle technology has increased the opportunities for the production of cost-effective and environmentally friendly fertilisers [11, 12]. They have a high potential to augment crop yield, due to increased nutrient use efficiency, in part by their slower release. This improved nutrient utilisation does not only reduce its loss but also controls environmental pollution [8].

Atar et al. [13] have also pointed out how beneficial iron oxide nanoparticles are to stimulate certain enzymes such as catalase, cytochrome oxidase and peroxidase. Kaviani et al. [14] reported that iron oxide nanoparticles played a supportive role in enhancing nutrient stress tolerance in plants since they provide cofactors that are constituents of many antioxidants.

Moreover, the high reduction potential of iron oxide nanoparticles makes them efficient to remove organic pollutants such as atrazine and chlorinated compounds from water [15, 16].

However, both positive and negative effects of iron oxide nanoparticles on plants have been reported [4, 8, 17,18,19]. Among the positive effects observed, after foliar application of nanoparticles, different authors have reported an improved photosynthetic rate, chlorophyll content, biomass, grain yield and nutritional quality in wheat [17], barley [20] and watermelon [21, 22]. Other authors have observed enhanced plant growth in clover [23] and soybean [24, 25] and also better nutrient uptake by increased microbial enzyme activity in the rhizosphere in rice [26], tomato [27], peanut [3], maize [28, 29] and pumpkin [30].

Iron is an essential nutrient for all organisms [31] and its deficiency is widespread among many different crops [32]. Low native soil fertility coupled with lower input use efficiency of nutrients is a persistent problem in many agricultural production systems [18]. Although the iron content in soils is usually high, a large proportion is fixed as insoluble Fe(III), especially in aerobic and high-pH soils [33]. These substrates are often deficient in Fe(II), which is the plant-available species and can lead to iron-deficient vegetation [34, 35].

In plants, iron is involved in many physiological processes, including chlorophyll biosynthesis, respiration and redox reactions [33, 34, 36]. Iron activates several enzymes, contributes to the ribonucleic acid synthesis and improves the performance of photosystems [18]. Moreover, most of the leaf iron (about 80%) is localised in the chloroplast, mainly in the molecular complexes involved in the photosynthetic electron transport chain, which contain about 60% of the total leaf iron [37, 38]. As a consequence, its deficiency mainly affects chloroplast structure and function. Regarding the food chain, iron deficiency not only has an adverse effect on plant growth and development but can also cause anaemia in animals and humans [39, 40]. The application of iron-containing fertilisers remains the most effective method to overcome this problem, hence the importance of improving the efficiency of iron fertiliser utilisation [3]. However, its excess may lead to toxic effects in plants through the generation of reactive oxygen species (ROS) and oxidative damage [41].

Magnetite nanoparticles (NP-Fe3O4, MAG), also known as superparamagnetic iron oxide nanoparticles (SPION), have applications in biological organisms other than plants. In fact, they can be used in various in vivo applications, such as magnetic resonance contrast enhancement, tissue repair, immunoassays, detoxification of biological fluids, hyperthermia, drug delivery and cell separation; thus, demonstrating that they are generally considered to be biologically innocuous [42,43,44]. All these biomedical applications take advantage of the magnetic properties of these small size nanoparticles. In addition to their use in the biomedical field [45,46,47,48], they have been increasingly employed for environmental remediation [49,50,51].

Within this context, the specific objective of this work was to determine the possible effects of Fe3O4 nanoparticles on the first steps of the plant photosynthetic chain, with a special focus on the electronic transport processes and their quantum yields. Considerable attention was given to the interaction of magnetite nanoparticles with photosystems and to the discussion of the possible mechanisms of action.

In particular, two main questions motivated this research:

  • Is there any beneficial or toxic effect, at the level of the initial stages of photosynthesis, which might condition the implementation of these nanomaterials as plant fertilisers or inoculant carriers?

  • Is there any relationship between this potential effect and the concentration of nanoparticles applied to the plant?

To answer these questions, we have sprayed leaves of Cichorium intybus with different concentrations of nanoparticles and we have applied several methodologies based on chlorophyll fluorescence analysis to infer information on the photosynthetic performance. We have complemented these measurements with reflectance spectra and determination of pigment content. The selected species (Cichorium intybus) is a worldwide-distributed plant, easily grown and cultivated and has the ability to tolerate a wide range of climatic and soil conditions. It is currently an important forage species in the Pampean plains (Argentina) and its leaves are large enough to carry out spectroscopic measurements with ease.

Results and discussion

Nanoparticles characterisation

Morphological analysis of magnetite nanoparticles

TEM image (Fig. 1A) reveals spherical, crystalline, well-resolved particles, displaying a narrow size distribution, with a mean diameter of 11 ± 2 nm.

Fig. 1
figure 1

Physicochemical characterisation of magnetite nanoparticles. A Transmission electron microscopy of magnetite nanoparticles. B Scanning electron microscopy of magnetite nanoparticles; inset: energy-dispersive X-ray spectroscopy of the nanoparticles. C X-ray diffraction spectrogram of magnetite nanoparticles. D Fourier-transform infrared (FTIR) spectrum. E Magnetisation hysteresis curve measured at 298 K. The blue curve represents fitting to the modified Langevin function. F Absorbance spectra for Fe3O4 nanoparticles dispersed in water. Numbers (1) and (2) represent samples from different batches

From multiple SEM images (Fig. 1B), a larger number of particles could be analysed using the Image J program. The diameter of almost five thousand particles in the range of 8 to 13 nm was so determined, with a median around 11 nm.

The energy-dispersive X-ray spectroscopy of the samples obtained by SEM–EDS (Fig. 1B) clearly evidenced the presence of a dominant phase of iron oxide. The ratio Fe/O obtained from their respective peaks was 39.1/48.7 = 0.8. Since the Fe/O atomic ratios in Fe3O4 and Fe2O3 are approximately 0.75 and 0.66, respectively, our result indicates that the predominant phase was actually magnetite (Fe3O4).

Structural analysis

The X-ray diffraction technique (XRD) patterns (Fig. 1C) indicated the presence of a single cubic spinel phase that belongs to the Fd3m space group. The main peaks corresponding to 2θ values: 30.2º (220), 35.6º (311), 43.3º (400), 53.7º (422), 57.3º (511) and 62.9º (440) agree with those for magnetite (JCPDS 19–0629) [52,53,54].

No other impurity phase was observed in the diffractogram (Fig. 1C), which indicates that the synthesized magnetite nanoparticles were high purity Fe3O4 particles with an inverse spinel structure. The strong and sharp peaks revealed that the Fe3O4 particles were highly crystallised, showing consistency with the morphologies predicted from the TEM images. The broadening of the diffraction peaks suggested a small size for the magnetite nanoparticles. In fact, the calculated crystal grain size of the magnetite nanoparticles, estimated from XRD patterns by means of the Scherrer equation, was 11.0 ± 0.8 nm, in agreement with TEM and SEM images (Fig. 1A and 1B).

From the Scherrer equation, the interplanar crystal spacing (d311) was calculated using the most intense peak (311). The distance d311 was 2.520 nm while the unit cell parameter (a) was 8.357 ± 0.003 Å. The cell parameter and interplanar spacing for the prepared sample were slightly lower than those reported for the bulk magnetite JCPDS standard (a = 8.394 Å and d311 = 2.531 nm). However, the values were similar to those reported in several references [10, 54,55,56]. The values obtained for the sample can be attributed to the simultaneous formation of another phase as γ-Fe2O3 (a = 8.342 Å and d311 = 2.517 nm, JCPDS No.39-1346) previously identified [55, 57]. The differences in unit cell parameters between our samples and pure magnetite indicate that the structure type can be defined as a defective spinel given the presence of a more compact impurity such as γ-Fe2O3 [58].

Fourier transform infrared analysis

FTIR spectrum for the nanoparticles from 4000 a 400 cm−1 is shown in Fig. 1.D. The intense bands at 569 and 638 cm−1 are attributed to Fe–O bonds [10, 59, 60], the position of the former being a function of nanoparticle size [60,61,62,63,64]. On the other hand, the band observed at 800 cm−1 could be associated with the Fe–O stretching mode of Fe2O3 [63]. This result suggests the presence of a small amount of Fe2O3 phase in the sample, which is consistent with that observed by X-ray diffraction.

The presence of O–H stretching at 3447 cm−1 and O–H bending at 1620 cm−1 is attributed to the coordinated OH groups or water molecules on the unsaturated surface of Fe atoms [64]. The C–O signals at 2365 and 2330 cm−1 are due to atmospheric CO2 [65]. The small absorption bands at 1410, 1550, 2850, 2950, 2920 and 2960 cm−1 are due to vibrations of NH4+ and NH3 residues that remained from the synthesis after washing and drying. Finally, the peaks at 1265 and 1100 cm−1 are consistent with the presence of either O–H and C–O groups from the solvent or solvent impurities [65].

Magnetic properties of nanoparticles

The magnetisation hysteresis curve measured at 298 K for Fe3O4 nanoparticles is shown in Fig. 1E. The good fitting of the curve to the Langevin equation and the lack of hysteresis (coercive field at M=0 and the remanent magnetization at H=0, both null) suggested that the nanoparticles exhibited dominant superparamagnetic behaviour together with a minor paramagnetic component.

By fitting this curve to the Langevin function, the saturation magnetisation (Ms = 69.9 ± 0.3 emu/g), the magnetic moment (µ = (2022 ± 34) µB, with µB = 9.27 J/T) and the paramagnetic component (χPM = 0.78 ± 0.03) were calculated. The obtained values were very close to those reported previously in literature for magnetite nanoparticles of similar size [10, 54, 64, 66,67,68].

The paramagnetic component is size-dependent; it usually increases with decreasing particle size. This has been associated with the surface spin disorder and uncompensated spins in nanoparticles of very small particle size [68].

Finally, hysteresis curves and the Langevin equation (see Experimental Section) enabled the estimation of a particle diameter of 14.0 ± 0.2 nm (usually referred to as the magnetic size), not identical but similar to the values obtained by electronic microscopy and X-ray diffraction.

Optical properties of nanoparticles

UV–VIS absorption spectra (Fig. 1F) were recorded and analysed to estimate the nanoparticles band gap (Egopt). The dried nanoparticles were dispersed in water for the measurements. Both sonicated and non-sonicated suspensions were essayed. Two absorption bands were present: one located at 240 nm and the other between 350 and 400 nm. The position and shape of the second band depended on whether the sample had been previously sonicated. Absorption was high in the UV and decreased gradually as the wavelength became longer.

Band gaps, obtained from Tauc plots, using absorbance spectra, were 3.8 ± 0.2 eV and 2.5 ± 0.2 eV, for direct transitions and 2.2 ± 0.2 eV and 1.4 ± 0.2 eV for indirect ones.

Analysis of hybrid systems: nanoparticles-leaves

Pigment content

Table 1 displays the results for pigment content and pigment ratio obtained for control leaves and for plants treated with magnetite nanoparticles.

Table 1 Pigment content (expressed in micrograms per gram of fresh leaf material) and pigment ratios for leaves treated with 0, 10, 100 and 1000 ppm magnetite nanoparticles

In general, a significant increase in the content of all pigments was observed in the presence of the nanoparticles. The Chl a/b ratio remained relatively constant, except for a sharp increase observed at 100 ppm. The ratio Chl/Car showed a significant increase for the treated plants. These results are similar to those reported for shaded leaves and can be ascribed to a shading effect, due to the absorption in the UV–Vis region of the nanoparticles. In fact, an increase in the Chls/Car ratio is usually associated with a lower level of the xanthophyll-cycle carotenoids, whenever shading conditions are present [69]. The xanthophyll cycle acts as a protective mechanism against high light intensities by providing an additional heat dissipation pathway. When the plant is acclimatised to low light intensities, this mechanism is usually downsized, since the plant no longer needs so much protection.

Additionally, an increase in the Chl a/b ratio might probably be related to an enhancement in the ratio photosystem I/II. This is because Chl a/b ratio in the light-harvesting complex I (LHC-I) is higher than for the LHC-II [70,71,72]. This result will be discussed later in combination with the spectral distribution of Chlorophyll fluorescence, recorded from the dark-adapted state.

Spectral reflectance indices

Reflectance spectra are shown in Fig. 2. A reduction of the mean reflectance in the visible part of the electromagnetic spectrum was observed, as magnetite concentration increased, with a reduction of around 25% at 550 nm for 1000 ppm. This decrease was associated with an increment in light absorption by the treated leaves, a fact that agrees with the presence of the particles and with higher pigment content.

Fig. 2
figure 2

Reflectance spectra of leaves treated with different concentrations of magnetite nanoparticles

In the NIR region, no significant changes were observed up to 100 ppm magnetite, while for 1000 ppm the reflectance decreased significantly. In this region, the absorption of leaf pigments is low, so reflectance is dominated by the internal structure of the leaf [73].

Regarding spectral indices, significant differences were only observed between the control plants and those treated with 1000 ppm magnetite (Table 2).

Table 2 Spectral indices calculated from reflectance spectra of control plants and plants treated with 1000 ppm magnetite nanoparticles

A strong agreement between these indices and pigment content was evident. In fact, the Normalised Difference Vegetation Index, NDVI, and the modified Normalised Difference Vegetation Index, mNDI, are related to chlorophylls concentration; the Pigment Specific Normalized Difference for Chlorophyll a, PSNDa, and the Pigment Specific Normalised Difference for Chlorophyll b, PSNDb, are proportional to concentrations of Chlorophyll a and b, respectively. Finally, the Photochemical Reflectance Index, PRI, is linked with foliar carotenoids. Therefore, increases in all these indices are in line with the higher pigment content in chicory leaves treated with 1000 ppm magnetite.

Fluorescence spectra

The results are shown here correspond to the spectral distribution of the initial Chl a fluorescence, which is the emission from dark-adapted leaves when excited by a low photon flux so as not to induce variable fluorescence or Kautsky’s kinetics (See experimental part).

A significant decrease in the intensity of the experimental spectra of initial fluorescence from treated leaves was observed, as the concentration of nanoparticles increased. The maximum decrease in fluorescence (recorded for 1000 ppm magnetite) was 36%. This trend still remained after spectra correction for light re-absorption processes.

The steady fluorescence spectra corrected for internal filter effects (light re-absorption) are shown in Fig. 3. The corrected fluorescence ratio Fred/Ffar-red decreased as the magnetite concentration increased (Table 3). A 14% reduction was observed for the 1000 ppm treatment.

Fig. 3
figure 3

Average initial fluorescence spectra of leaves treated with different concentrations of magnetite nanoparticles, corrected for the detector response to wavelengths and for light re-absorption processes. The average was carried out for 6 leaves of each treatment. Excitation wavelength: 460 nm

Table 3 Corrected Fred/Ffar-red ratio for leaves treated with different concentrations of magnetite nanoparticles

A reduction in the fluorescence ratio, once corrected by the inner filter, together with the increase in the ratio Chl a/ b obtained for the samples treated with the nanoparticles, might suggest a decrease in PSII relative to PSI or to an increase in PSI relative to PSII [69]. Nevertheless, this would need further validation, since a complex set of biological processes that determines the value of the chlorophylls ratio exists [74].

OJIP transient

As it can be observed in Fig. 4, chlorophyll fluorescence transients underwent substantial changes in response to magnetite treatment. The intensity of the curves decreased significantly after 7 days of treatment, displaying fluorescence quenching along the whole curve (34% for F0 and 19% for FM). The reduction of F0 was consistent with results observed for the steady-state fluorescence spectra for the dark-adapted state, in conditions where there was no induction of Kautsky’s kinetics (Section “Fluorescence spectra”).

Fig. 4
figure 4

OJIP Analysis. A) Average OJIP transients on a semi-logarithmic scale after 7 days of treatment. B) Variable fluorescence normalised in O and P (VOP = (Ft − F0) / (FM − F0)) and difference with control (ΔVOP = VOP, sample − VOP, control) as a function of time. C) Variable fluorescence normalised in O and J (VOJ = (Ft − F0) / (FJ − F0)) and difference with control (ΔVOJ = VOJ, sample—VOJ, control) as a function of time. D) Variable fluorescence normalised in O and K (VOK = (Ft − F0) / (FK − F0)) and difference with control (ΔVOK = VOK, sampleVOK, control) as a function of time. The curves represent the average of 10–20 measurements (obtained from three plants per treatment)

The decrease in fluorescence along the entire OJIP transient, for magnetite-treated leaves, led to a bigger area above the curve. As this area is directly related to the number of electron carriers, the first straight information is that this number is enhanced in the presence of magnetite. In fact, after treatment with nanoparticles, both the normalised area (SM) and the total number of transferred electrons (N) increased significantly (Table S2 in the Supplementary Material).

Because different inflection points in the OJIP curve represent different stages of the electronic transport, calculating the variable fluorescence for each sample and then plotting the difference from the control, allowed the identification of the main changes in the OJIP transient. In fact, differences between control and treated leaves were more evident in the normalised difference plots than in the original OJIP transients (Compare Fig. 4B, C and D with A).

A pronounced negative band ΔVOJ was observed at 2–3 ms (Fig. 4B). This region is highly dependent on the redox state of the primary electron carriers, mainly QA/QA on the acceptor side of PSII, and this is related to the probability of an electron advancing in inter-system electron transport towards PSI [75]. This means that a positive ΔVOJ band indicates an accumulation of reduced carriers such as PQs, Cyt-b6f and PCs, whereas a negative ΔVOJ band points to the opposite effect, which can be interpreted as an increase in PSI activity.

In addition, a negative ΔVOI band was observed at approximately 30 ms (Fig. 4B). A positive ΔVOI band suggested the inhibition of the reduction of terminal electron acceptors, such as Fdred and NADP+, or lower activity of PSI [76]. Then, the negative ΔVOI band observed in our experiments indicated an opposite effect and was consistent with an increase in PSI activity [76].

Differences for curves normalised in O and J (ΔVOJ) and in O and K (ΔVOK) are displayed in Figs. 4C and D, respectively.

A negative K peak, placed at 0.3 ms, appeared in ΔVOJ curves. The presence of a positive K peak is usually related to the inactivation of the oxygen-evolving complex (OEC), an increase in the size of LHC-II or to changes in the redox state of the plastoquinone pool [77, 78]. Damage in the OEC is usually inferred from the ratio FK/FJ, which indicates the inactivation degree of PSII donor side [79] or from a decrease in the active fraction of the OEC calculated from VK and VJ values. In our research, the observed K peak was negative, FK/FJ decreased and the active fraction of OEC increased in the presence of nanoparticles. These results indicate that the participation of OEC in electron donation to PSII is enhanced for leaves with incorporated nanoparticles.

A negative L peak placed at 120 µs appeared in ΔVOK curves. The L-band, which typically appears at 0.12–0.15 ms, increases when connectivity or groupings in PSII units are lowered [76, 80, 81]. A negative value for the L-band observed in our experiments suggested that the grouping of PSII units was enhanced in the presence of nanoparticles.

The application of the model initially proposed by Duysens and Sweers in 1963 [82] made possible the calculation of the efficiencies and yields related to the energy fluxes of each stage of the electronic transport for the different treatments. The results are presented in Fig. 5 and Table S2. For more information on these parameters see Table S1 in the Supplementary Material.

Fig. 5
figure 5

Radial distribution plot of selected OJIP parameters for leaves treated with different concentrations of magnetite nanoparticles. For each parameter, the value of the control was set to 1

Values for yields and efficiencies of electronic transport were higher for treated leaves than for control plants in all cases. The maximum change of these parameters (corresponding to a dose of 100 ppm) were: for the maximum yield of primary photochemistry (ΦP0), + 9%; for the quantum yield of electronic transport flux from QA to QBET0), + 45%; for the probability by which a trapped electron in PSII is transferred from QA to QBET0), + 36%; for the probability by which an electron is transferred from QB up to PSI acceptors (δRE), + 23%; for the probability by which a trapped electron in PSII is transferred up to PSI acceptors (ΦRE0), + 100%; and for the quantum yield of electronic transport flux up to PSI acceptors (ΨRE0), + 80%.

Even though the whole set of yields and efficiencies increased in the presence of the nanoparticles, change magnitudes were vastly different for each stage, as the obtained values show.

The important variation for ΨET0, which represented an increase in electron transfer from QA to QB, corresponded to the inflection point J, where a local minimum appeared when plotting ΔVOP curves in Fig. 4B. Moreover, the very remarkable rise in ΦRE0 indicated an improvement in PSI activity, which was also in agreement with the negative band ΔVOI observed at 30 ms (Fig. 4B).

The absorption flux per reaction center of active PSII (ABS/RC) changed up to − 36%; the trapped energy flux per reaction center (TR0/RC) varied up to − 36% and the dissipated excitation energy flux per reaction center (DI0/RC) changed up to − 48% (Fig. 5). Additionally, the flux of electron transport between QA and QB per reaction centre (ET0/RC) augmented up to + 8% (100 ppm dose) and the electronic transport flux to PSI acceptors per active reaction center of PSII (RE0/RC) increased up to 37%.

Despite the decrease in absorption and trapping fluxes, the electronic transport fluxes ET0/RC and RE0/RC increased significantly (Fig. 5).

The results indicate that the flux of electron transport from QA to QB increased, even though the flux of trapped excitons declined. The appearance of a negative ΔVOJ band also reflected an increase in the conversion of QA to QA.

The enhancement of the activity of the PSI activity, initially proposed by the observation of a negative ΔVOI band, was also supported by the increase in fluxes and yields associated with PSI such as RE0/RC, δRE, ΦRE0 and ΨRE0.

Finally, γRC (probability that chlorophyll in PSII acts as a reaction center), PIABS and \({\mathrm{PI}}_{\mathrm{total}}^{\mathrm{ABS}}\) increased in the presence of nanoparticles. The parameter PIABS quantifies the energy conservation of the photons absorbed by PSII up to the reduction of the intermediate electron acceptors between PSII and PSI, whereas \({\mathrm{PI}}_{\mathrm{total}}^{\mathrm{ABS}}\) extends the energy conversion up to the PSI acceptors. \({\mathrm{PI}}_{\mathrm{total}}^{\mathrm{ABS}}\) is considered a very appropriate proxy for the overall health of the photosynthetic unit [83].

In summary, based on a detailed analysis of the curves and parameters derived from the OJIP transient, effects at different points of the photosynthetic electronic transport chain have been determined. The parameters derived from the absorption and primary charge separation suggest that one of the effects of the particles is the competitive absorption of the incident light and the fluorescence quenching, which leads to reduced values for F0 and FM, as well as in the flux of absorbed photons per active reaction centre (ABS/RC). The latter change could be also due to an increase in active reaction centres.

The analysis of electron transport fluxes and yields suggested an increase in the fluxes associated with electron transfer from QA to the PSI acceptors.

In particular, the hypothesis of a possible increase in PSI activity is consistent with multiple changes observed in the ΔVOI and ΔVOJ bands, and in the δRE and RE0/RC values. All these changes agree with the increase in the total number of transferred electrons (N) and the normalised area over the curve (SM).

Several interesting papers have shed light on the effect of different nanoparticles types on photosynthetic genes that in turn influence global photosynthetic activity [20, 84, 85]. Our results, together with bibliographic information, highlight a close connection between Fe3O4 nanoparticles and photosystems, especially with PSI. In fact, recent studies showed that nanoparticles, similar in size and composition to those presented in this work, upregulated the photosystem I P700 chlorophyll a apoprotein A1 (PsaA), which is a photosynthetic gene encoding subunits of the PSI reaction centre [20]. The nanoparticles also upregulated the gene PetA which encodes cytochrome b6f complexes. Stimulation of Cyt-b6f will result in an improvement of the electron transport between PSII and PSI and agrees with the negative value obtained here for ΔVOJ band. In addition, the enhancement of PSI activity is consistent with an efficient electron capture by this photosystem and explains the negative value for ΔVOI band obtained in this work. Overall, the stimulated induction of these photosynthetic genes by Fe3O4 nanoparticles is in line with the increments observed for the probability by which trapped electrons in PSII are transferred up to PSI acceptors and also for the important rise in \({\mathrm{PI}}_{\mathrm{total}}^{\mathrm{ABS}}.\)

Energy partition: actinic phase

From Kautsky’s kinetics, a series of photosynthetic parameters were obtained. The results are summarised in Fig. 6 and Table S3. The initial fluorescence F0 decreased in the presence of nanoparticles, consistently with the results previously shown in this work for the OJIP curve and the steady-state fluorescence spectra. The ratio FV/FM, corresponding to the maximum photochemical quantum yield of PSII, augmented with an increasing concentration of magnetite nanoparticles. Both FV/FM values and their variations in the presence of nanoparticles are consistent with those observed for its equivalent parameter ΦP0, which was analysed previously from the OJIP curves.

Fig. 6
figure 6

Derived parameters from Kautsky kinetics under actinic illumination for leaves treated with nanoparticles and for control plants. The values between the different treatments marked by the same letter (ad) were not significantly different at p < 0.05. In this figure, FV/FM represents the maximum photochemical quantum yield of PSII, ΦPSII is the quantum yield of PSII, ΦNPQ is the quantum yield of non-photochemical quenching, qp is the coefficient of photochemical quenching and qNp is the coefficient of non-photochemical quenching

Energy partition during actinic illumination among photosynthesis (ΦPSII), heat dissipation by non-photochemical quenching (ΦNPQ) and photophysical decay including fluorescence (Φc) was also analysed. The parameter ΦPSII augmented in the presence of nanoparticles with a maximum increase of 86% at 100 ppm magnetite. The nanoparticles induced a decrease in ΦNPQ (-31%, maximum for 100 ppm magnetite) and a slight decline for Φc.

Reduction in NPQ was completely consistent with the drop in carotenoid content as these pigments are present in the xanthophyll cycle directly involved in NPQ processes.

The coefficient for photochemical quenching (qP), which represents the events increasing the consumption of photosynthetic electrons, augmented upon nanoparticle treatment reaching a maximum value for 100 ppm magnetite.

Finally, the photosynthetic rate of PSII (PSrate), which is obtained from the product of the photosynthetic yield of PSII and the intensity of the actinic light [69], also increased in magnetite presence.

These results, all together, pointed to the fact that nanoparticles favoured photosynthesis in the distribution of absorbed energy. In this study, an optimum concentration of nanoparticles (100 ppm) was present for the benefits of the photosynthetic apparatus, above which the positive effects were lower.

Conclusion

Treatment with the synthesised magnetite nanoparticles improved photosynthetic parameters in chicory plants. The beneficial effect was observed progressively, with the applied concentration of magnetite up to 100 ppm. A further increase to 1000 ppm of magnetite resulted in a decline of the parameters in relation to 100 ppm but not in comparison with the control. This indicates that there is an optimum concentration of magnetite to stimulate photosynthesis, above which phytotoxic effects may appear.

While the most striking result was the enlargement of PSI activity with respect to PSII, the whole electronic transport chain increased its activity in the presence of the nanoparticles. This observation is consistent with the occurrence of the interactions nanoparticles-photosynthetic complexes. Comparing our results with other research works, we can conclude that these interactions are probably established by means of up-regulation of photosynthetic genes.

As a final reflection, their potential application in crop management is worth analysing. It is well documented in the literature that plants require Fe (II) for their biochemical processes, but ferrous oxide is an unstable species. Additionally, several of the reported negative effects of iron-nanoparticle application on plants are caused by the presence of Fe2O3 [86]. Magnetite, Fe3O4, is a stable form of iron compound able to provide Fe (II) to plants [87]. This, supported by the results obtained in this work, can conclude that the magnetite nanoparticles synthesised in this research, containing a very low content of Fe2O3 phase, constitute a promising material to be used as a fertiliser and should be further investigated for this application.

Experimental section

Fe3O4 nanoparticles synthesis

Fe3O4 nanoparticles were synthesised by co-precipitation of Fe3+ and Fe2+ by slightly modifying the protocol previously reported by Sun and Zeng [88]. Briefly, 6.16 g of analytical grade iron(III) chloride (FeCl3.6H2O) and 3 g of iron(II) chloride (FeCl2.4H2O) were dissolved in 100 ml of double-distilled water with vigorous stirring at 90°C. Then, 10 ml of 25% NH4OH was gradually added until a pH of about 10 was reached. Successful precipitation of Fe3O4 was evidenced by a colour change in the suspension from brown to black, which occurred rapidly after the total addition of ammonia. After a further 30 min of stirring at a constant temperature, the suspension was allowed to cool to room temperature. The magnetic particles obtained were repeatedly washed with distilled water using a magnet to facilitate their settling. Once the supernatant was completely translucent, the magnetic solid was left to dry for 24 h at around 70 °C, with the caution not to exceed this temperature. Subsequently, the solid obtained was stored in a glass bottle within a desiccator until use.

Electron microscopy

Scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) images were obtained with a Zeiss Supra 40 microscope. A voltage between 5 and 10 kV (depending on the sample) and magnifications between 100 and 800 kX were applied. ImageJ image processing software was used for size estimation from SEM images, by counting a total of 4758 particles from several SEM images taken on the same day for the same sample.

Transmission electron microscopy (TEM) images were captured with a Phillips EM 301microscope, operated at 60 kV with a resolution in the order of 5 nm. Images were obtained at magnifications between 25 and 100 kX. The images were analysed in the same way as the SEM micrographs.

X-ray diffraction

To detect the presence of crystalline phases in the synthesised materials, X-ray diffraction (XRD) in grazing incidence mode was used. Measurements were carried out with a Siemens D5000 diffractometer, using Cu radiation source K = 1.54056 Å. The diffractograms were recorded in an interval from 15 to 70 degrees, with a reading step of 0.026 and a counting time of 22 s per step.

Magnetic hysteresis cycles

The magnetic hysteresis cycles were measured with a LakeShore 7400 vibrating sample magnetometer (VSM). A small amount in the order of 50 mg of magnetic sample powder was weighed and wrapped in Teflon tape to form a small packet, which was then placed in the sample holder of the instrument. The magnetisation-demagnetisation cycles were measured at room temperature, varying the induced field between -10 kOe and 10 kOe in steps of 500 Oe, with a signal integration time of 10 s for each applied magnetic field.

Then, the field-dependent magnetisation curves were fitted using the Langevin equation for superparamagnetic particles [89, 90], to obtain parameters of interest such as the magnetic moment of the particles, the saturation magnetisation and the magnetic size of the particles [89, 91].

Optical properties of Fe3O4 nanoparticles

The absorption spectra and band gaps of magnetite nanoparticles (MAG) were determined using a UV–Vis spectrometer (UV-3600 Plus, Shimadzu, Tokyo, Japan), from 200 to 800 nm. The nanoparticles were ultrasonically dispersed in deionised water, to obtain a uniform dispersion of the nanoparticles prior to measurement. Low nanoparticle concentrations, in the order of 10 ppm, were used for a correct spectroscopic determination.

From absorption spectra, the band gap may be calculated by using the Tauc relation [92, 93]. This relation connects the absorption coefficient α with the optical band-gap (Eg), as described in Eq. (1).

$$\alpha h\nu =B.{\left(h\nu -{E}_{g}\right)}^{n}$$
(1)

where B is a constant, is the photon energy and n is a value dependent on the transition type (n = 2 for allowed indirect transitions, 3/2 for forbidden direct transitions and ½ for allowed direct transitions). A plot of (αhν)1/n vs hν and the subsequent extrapolation of the linear region to the abscissa yields the energy of the optical bandgap (notice that when α = 0, Eg = ).

The coefficient α may be calculated from Eq. (2) [94]:

$$\alpha =2.303.{10}^{3}\frac{\rho .A}{l.c.M}$$
(2)

where, A is the absorbance, ρ is Fe3O4 bulk density (5.18 g.cm−3), l is the optical pathway, c is the molar concentration and M, the molecular weight.

By using Eqs. (1) and (2) combined with the experimental absorbance spectra, optical band-gaps for the allowed direct transition (Eg1opt) and for the allowed indirect transition (Eg2opt) could be calculated.

Plant growth and nanoparticles application

Chicory (Cichorium intybus) seeds were planted and, when the sprouts reached a suitable size, they were transplanted individually into pots. The plants were kept under controlled irrigation and illumination until they had a suitable size and their photosynthetic parameters reached an optimal and constant value. The plants were acclimatised to medium irradiance (300 µmol photons m−2 s−1), with a photoperiod of 12 h for 30 days before the treatment with nanoparticles started.

Dispersions of 10, 100 and 1000 ppm of magnetite nanoparticles in distilled water were prepared and 25 mL were sprayed on the plant leaves for 7 days. This methodology had been previously studied and proved effective for the incorporation of nanomaterials into leaves [21]. Control plants were sprayed with distilled water instead of nanoparticle suspension. Each treatment included 4 plants of a similar number and size of leaves.

Pigment content

Leaf samples were washed with distilled water and the veins and petioles were removed. Photosynthetic plant pigments, chlorophylls and carotenes, were extracted with 80% acetone, using a mortar and pestle. The extracts were centrifuged for 3–5 min in glass tubes until the supernatant was completely clear. The absorbance spectrum was then recorded in the visible range at an appropriate dilution. A double beam spectrophotometer with quartz cuvettes (UV-3600 Plus, Shimadzu, Tokyo, Japan) was used. The specific absorption coefficients of chlorophyll a (Chl a), chlorophyll b (Chl b) and total carotenoids (Cars) reported by Lichtenthaler and Buschmann (2005) [95] were used for calculations. The weight ratios of the pigments, Chl a/b and Chls/Cars, were then determined. Pigment determination was carried out on leaves removed from the plants on the seventh day of treatment.

Spectral reflectance indices

Diffuse reflectance spectra (R(λ)) as a function of wavelength (λ) of control and treated leaves were recorded using a spectrophotometer (3101PC, Shimadzu, Tokyo, Japan) equipped with an integrating sphere. A reference of Barium sulphate was used to set 100% reflectance. From the reflectance values at different wavelengths, spectral indices were calculated from Eqs. (3) to (7):

$$\mathrm{NDVI}= \frac{{R}_{800 \mathrm{nm}}- {R}_{680 \mathrm{nm}}}{{R}_{800 \mathrm{nm}}+ {R}_{680 \mathrm{nm}}}$$
(3)
$$\mathrm{mNDI}= \frac{{R}_{750 \mathrm{nm}}- {R}_{705 \mathrm{nm}}}{{R}_{750 \mathrm{nm}}+ {R}_{705 \mathrm{nm}}-2 {R}_{445 \mathrm{nm}}}$$
(4)
$$\mathrm{PRI}= \frac{{R}_{531 \mathrm{nm}}- {R}_{570 \mathrm{nm}}}{{R}_{531 \mathrm{nm}}+ {R}_{570 \mathrm{nm}}}$$
(5)
$${\mathrm{PSND}}_{a}= \frac{{R}_{800 \mathrm{nm}}- {R}_{675 \mathrm{nm}}}{{R}_{800 \mathrm{nm}}+ {R}_{675 \mathrm{nm}}}$$
(6)
$${\mathrm{PSND}}_{b}= \frac{{R}_{800 \mathrm{nm}}- {R}_{650 \mathrm{nm}}}{{R}_{800 \mathrm{nm}}+ {R}_{650 \mathrm{nm}}}$$
(7)

Spectral distribution of initial fluorescence: dark-adapted state

The spectral distribution of the initial chlorophyll fluorescence (F0) was obtained on the adaxial side of the leaves by using a steady-state spectrofluorometer in front-face geometry with an angle of 60° (QuantaMaster, PTI-Brunswick, USA). Spectra were recorded at room temperature on freshly cut leaves which had been dark-adapted for 20 min. The excitation wavelength was set at 460 nm, and spectra were recorded between 600 and 800 nm. They were corrected by the response of the detector at each wavelength. To avoid inducing Kautsky’s kinetics, a photon flux lower than 20 µmol m−2 s−1 for the excitation beam was used.

Then, the stationary fluorescence spectra were corrected according to Ramos and Lagorio [96], to eliminate the distortion produced by light re-absorption processes. For this purpose, the diffuse reflectance spectra of an optically thick layer of leaves were recorded using a spectrophotometer (3101PC, Shimadzu, Tokyo, Japan) equipped with an integrating sphere. Barium sulphate was used for 100% reflectance adjustment. Measurements were carried out on the seventh day of treatment.

Variable Chlorophyll fluorescence. OJIP transient

Chlorophyll fluorescence transients (OJIP) were recorded with a Plant Efficiency Analyser (Handy-PEA, Hansatech Instruments Ltd., UK). The OJIP curve was induced using a saturating red light of 3000 µmol m−2 s−1 photon. The determination was performed on intact dark-adapted leaves. OJIP curves were recorded for an average of 20 leaves per treatment. The quantification of fast fluorescence transients was performed according to the assumptions and flux model previously described in reference [97]. Experiments were carried out on days: 1, 3, 5 and 7 from the start of treatment.

Variable chlorophyll fluorescence: Kautsky’s kinetics

The variable fluorescence was analysed using a pulse-modulated fluorometer (Hansatech Instruments, FMS1). Details on this technique and on the photosynthetical derived parameters can be found in references [98, 99].

The plant leaves were adapted to the dark for 20 min to record the initial fluorescence F0. Then, a saturating pulse was applied, and a maximum value for the fluorescence was attained (FM). Afterwards, a medium actinic light was turned on and the fluorescence was allowed to stabilise to a stationary value (FS). Once this point was reached, a new saturating light pulse was applied to obtain FM′.

This fluorometer had two different light sources, a sample illumination beam, which could be either saturating or actinic, and a modulated measurement beam of low intensity that did not induce variable fluorescence. The modulated beam (594 nm), which used pulses of very short duration (1.8 s) with long periods of inactivity between pulses, induced a modulated fluorescence signal. In this way, the integrated amount of radiation incident on the sample was less than 0.05 µmol photons m−2 s−1, avoiding significant physiological changes in the sample during measurements. The saturation pulse (halogen light) had a duration of 0.5 s with an intensity of 2700 µmol photons m−2 s−1. The actinic light was provided by the same halogen light source, with an intensity varying between 0 and 1039 µmol photons m−2 s−1.

Experiments were carried out on days: 1, 3, 5 and 7 from the start of the treatment.

From the experimental signals recorded, the following parameters connected with the photosynthetic performance of plants were calculated: the maximum quantum yield of photosynthesis for dark-adapted leaves, FM − F0/FM; the quantum yield of photophysical decay, Φc (Fs/FM); the quantum yield of non-photochemical quenching, ΦNPQ (Fs (FM − FM′)/(FM FM′)); the efficiency of PSII, ΦPSII (FM′ − FS/FM′); the photosynthetic rate of PSII, PSrate (ΦPSII. AL, with Al: intensity of the actinic light); the coefficient for the photochemical quenching, qP, ((FM′ − Fs)/(FM′ − F0)) and the coefficient for non-photochemical quenching, qNP ((FM − FM′)/(FM − F0)). References [98,99,100] provide more details.

Statistics

Data processing was carried out in Python by means of the Numpy module and Pandas. The statistical analysis was performed using one-way ANOVA with a Tukey HSD post-hoc test to determine if values from different groups were significantly different (Pingouin package was used).