crystals

Article Controlled Synthesis of Magnetic Iron Oxide Nanoparticles: Magnetite or Maghemite?

Sebastian P. Schwaminger * , Christopher Syhr and Sonja Berensmeier *

Bioseparation Engineering Group, Department of Mechanical Engineering, Technical University of Munich, 85748 Garching, Germany; [email protected] * Correspondence: [email protected] (S.P.S.); [email protected]; (S.B.); Tel.: +49-89-2891-5747 (S.P.S.); +49-89-2891-5750 (S.B.)


Received: 2 March 2020; Accepted: 13 March 2020; Published: 19 March 2020


Abstract: Today, magnetic nanoparticles are present in multiple medical and industrial applications. We take a closer look at the synthesis of magnetic iron oxide nanoparticles through the co-precipitation of iron salts in an alkaline environment. The variation of the synthesis parameters (ion concentration, temperature, stirring rate, reaction time and dosing rate) change the structure and diameter of the nanoparticles. Magnetic iron oxide nanoparticles are characterized by X-ray diffraction (XRD), Raman spectroscopy and transmission electron microscopy (TEM). Magnetic nanoparticles ranging from 5 to 16 nm in diameter were synthesized and their chemical structure was identified. Due to the evaluation of Raman spectra, TEM and XRD, the magnetite and maghemite nanoparticles can be observed and the proportion of phases and the particle size can be related to the synthesis conditions. We want to highlight the use of Raman active modes A1g of spinel structured iron oxides to determine the content of magnetite and maghemite in our samples. Magnetite nanoparticles can be derived from highly alkaline conditions even without establishing an inert atmosphere during the synthesis. The correlation between the particle properties and the various parameters of the synthesis was modelled with linear mixture models. The two models can predict the particle size and the oxidation state of the synthesized nanoparticles, respectively. The modeling of synthesis parameters not only helps to improve synthesis conditions for iron oxide nanoparticles but to understand crystallization of nanomaterials.

Keywords: iron oxide nanoparticles; co-precipitation; synthesis; magnetic nanoparticles; X-ray diffraction; raman spectroscopy; transmission electron microscopy; design of experiments

1. Introduction Nanotechnology is of great interest to multiple research fields due to the change of physical and chemical properties when reducing the dimension of a material to 1–100 nm. There are multiple fields of applications for these fascinating little particles ranging from packing and textile industry to catalysis and semiconductor industry [1]. While often the surface to volume ratio, the reactivity, optical or electrical properties when using nanomaterials can be employed in applications, iron oxides possess superparamagnetic properties at the nanoscale [2–4]. The magnetic properties are temperature dependent since the Brownian motion and the Neél relaxation, which lead to a fluctuation of magnetic fields on single nanoparticles are temperature dependent [4]. For iron oxides (magnetite and maghemite), the critical particle diameter leading to no remanence is below 20–30 nm at room temperature (300 K) [3]. With these magnetic properties, iron oxide nanoparticles are the basis for multiple applications such as magnetic imaging, magnetic targeting and hyperthermia [4,5]. Aside from medical applications, iron oxide nanoparticles can be used in protein purification [6,7], enzyme and catalyst immobilization [8–10], sensing [11] and energy storage [12,13].

Crystals 2020, 10, 214; doi:10.3390/cryst10030214 www.mdpi.com/journal/crystals Crystals 2020, 10, 214 2 of 12

For these applications, an understanding and development of effective and low-cost synthesis routes are necessary. The size, stability and composition of magnetic nanoparticles need to be controlled in order to affect magnetic, optic and surface properties [5,14]. Since these physical and chemical properties are often related to the dimension of nanomaterials, the most simple approach is to investigate the size of nanoparticles for distinct synthesis conditions [2,15–17]. Forge et al. and Roth et al. investigated the influence of several parameters on the particle size of iron oxide nanoparticles based on the alkaline co-precipitation synthesis [17–19]. Other studies focus on the formation pathways of iron oxide nanoparticles depending on the synthesis conditions and especially on the reaction time [20]. Such studies include the formation, which can be quite versatile for multiple synthesis routes [2,14,20–26]. Although some works approach parameter studies by design of experiments, multiple questions concerning the co-precipitation are not understood [17,18]. For these nanoscale particles, it is quite challenging to characterize and predict their structure, especially at the surface [14,27]. Multiple approaches exist on classifying magnetic iron oxide nanoparticles and describing them according to their properties, which are strongly dependent on their environment, synthesis conditions, stabilizers, and their age [4,28]. A further issue is the oxidation and phase transformations occurring during synthesis as well as during storage and use of iron oxide nanoparticles [28]. The oxidation mechanism and the transition of ions within the nanoparticles and lattice distortion are not completely understood due to the complexity and size-dependence of nanoscale systems [28–30]. Thus, multiple questions arise: In what phase are those nanoparticles really ordered? How strongly dependent is this structure on the particle size? How can results obtained by different techniques such as Raman spectroscopy, Mössbauer spectroscopy, magnetic measurements or XRD be interpreted, since size, surface and phase effects overlay? We want to correlate the synthesis parameters of a cost-effective co-precipitation synthesis route with the properties of the magnetic nanoparticles. Therefore, the systematic parameter variation is analyzed with design of experiments (DoE) and the parameters of educt concentrations (iron salts and sodium hydroxide), reaction time and temperature as well as dosing and stirring rate are evaluated towards their influence on particle size and composition. The particle size is analyzed with transmission electron microscopy (TEM) and X-ray diffraction (XRD) where the lattice disorder can be correlated to the particle size via the Scherrer equation. The lattice constant, obtained from the Bragg equation from the position of reflections in XRD, is correlated to the content of magnetite and maghemite in the synthesized nanoparticles [31]. The magnetite content is further investigated with Raman spectroscopy as a better measure for the ratio of Fe3O4 to γ-Fe2O3, due to different vibration modes corresponding to different distances between Fe–O for the spinel and distorted spinel structure. We are the first to use Raman spectroscopy to statistically evaluate the structure of iron oxide nanoparticles synthesized with co-precipitation over a wide parameter range. With these data, we are able to correlate the synthesis parameters to the iron oxide phase of obtained nanoparticles, which can be very useful for multiple applications.

2. Materials and Methods

2.1. Design of Experiments Design of experiments was planned with the software “Design Expert”. The defined parameters and their range is illustrated in Table1. All conditions for each experiment are summarized in Tables S1 and S2.

Table 1. Parameter range for design of experiments. For the substrates, the parameter range is between 0 und 1.

Parameter 1 0 1 − A1: Iron(II)chloride in g 3.5 7 A2: Iron(III)chloride in g 8.64 17.28 B: Sodium hydroxide in g 7.2 14.4 C: Temperature T in ◦C 30 55 80 D: Stirring speed in rpm 0 500 1000 E: Titration time in minutes 0 30 60 F: Reaction time in minutes 0 60 120 Crystals 2020, 10, 214 3 of 12

2.2. Synthesis

Ferrous and ferric chloride were used as received from Sigma-Aldrich as Fe(III)Cl3x(H2O)6 and Fe(II)Cl2x(H2O)4, respectively. In accordance with Table1, the salts were weighed and dissolved in 20 mL of deionized and degassed water. Sodium hydroxide (Fluka) was dissolved in 100 mL of deionized and degassed water, leading to a total reaction volume of 140 mL for each experiment. The sodium hydroxide solution was stirred with a magnetic stirrer in a glass beaker and the reaction temperature was adjusted before adding the iron chloride solutions with the respective dosing speeds. For long reaction times at high temperatures, the reaction volume was kept constant by adding deionized and degassed water. The co-precipitated product was magnetically decanted and washed with deionized and degassed water until a neutral pH (7–8) and a low ionic strength (<200 µS/cm) was reached. For XRD and Raman spectroscopy analysis, the samples were freeze-dried.

2.3. Characterization For TEM analysis, the samples were diluted and precipitated on carbon-coated copper grids. A JEOL JEM-1400 Plus (JEOL (Germany), Freising, Germany) was used for the microscopic characterization of nanoparticles. For each sample at least five pictures were taken and analyzed with ImageJ. Thereby, a minimum of 30 particles were counted for each picture. XRD experiments were carried out in transmission geometry of powder samples with a STOE Stadi P diffractometer (STOE & Cie GmbH, Darmstadt, Germany) using a molybdenum Kα source (λ = 0.709 Å). The range of 2–48◦ was analyzed with a Dectris detector (DECTRIS Ltd., Baden-Daettwil, Switzerland). Particle size was estimated from reflex broadening according to the Scherrer equation with a K factor of 0.89. The 440 reflection was fitted with a Voigt function with the software Origin in order to analyze the content of magnetite and maghemite based on the areas (A) of the respective reflections considering the following Equation (1).

A(Magn440) Magnetite (%) = 100 ( ) (1) A(Magn440) + A(Magh440)

The Raman spectroscopy of powder samples was performed with a Raman SENTERRA from Bruker (Bruker Optics, Ettlingen, Germany). A 488 nm laser was used at 0.1 mW power for 30 seconds. Two co-additions were used to correct cosmical spikes. All measurements were performed in triplicates. The spectra were further processed with a concave rubber band correction in the software OPUS. 1 1 The bands at 660 cm− and 710 cm− were fitted with Voigt functions in Origin and used for magnetite content analysis (OriginLab Corporation, Wellesley, Massachusetts, US). The deconvolution was processed with Equation (2) in order to obtain the magnetite share [28].

A(710) Magnetite (%) = 100(1 ) (2) − A(660)

3. Results The analysis of TEM and XRD results yield the average sizes of synthesized iron oxide nanoparticles, which range between 5 and 16 nm. Generally, the TEM results indicate a slightly larger diameter than the XRD diameters of around 0.3 nm, which is consistent with literature. The results of all experiments are shown in Figure1a and Table S3. A typical di ffractogram of magnetic iron oxide nanoparticles is shown in Figure1b. The full width at half maximum of the 311 reflection was used for the size analysis with the Scherrer equation, while the 440 reflection was evaluated towards phase analysis and discrimination between magnetite and maghemite [31]. Crystals 2020, 10, 214 4 of 12 Crystals 2020, 10, x FOR PEER REVIEW 4 of 12 Crystals 2020, 10, x FOR PEER REVIEW 4 of 12

FigureFigure 1.1. The diameters obtained from TEM and XRD analysis are compared for each experiment ((a)a).. Figure 1. The diameters obtained from TEM and XRD analysis are compared for each experiment (a). DiDiffractogramffractogram of of sample sample 21, 21, which which illustrates illustrates the the reflexes reflexes which which can can be found be found for spinel for spinel structures structures such Diffractogram of sample 21, which illustrates the reflexes which can be found for spinel structures assuch Fe 3asO 4Feand3O4 γand-Fe 2γO-Fe3 and2O3 aand reflection a reflection broadening broadening corresponding corresponding to 9.3 to nm 9.3 (nmb). The(b). The reflections reflections are such as Fe3O4 and γ-Fe2O3 and a reflection broadening corresponding to 9.3 nm (b). The reflections indicatedare indicated with with Miller Miller indices. indices. are indicated with Miller indices. WhileWhile the the XRD XRD data data represents represents a statistical a statistical representation representation of lattice errors of lattice and crystal errors deformations and crystal While the XRD data represents a statistical representation of lattice errors and crystal duedeformations to the nanoscale due to the confinement nanoscale confinement of particles, of the particles, TEM analysisthe TEM allowsanalysis a allows shape a analysis shape analysis and a deformations due to the nanoscale confinement of particles, the TEM analysis allows a shape analysis rudimentaryand a rudimentary analysis analysis of the particleof the particle distribution. distribution. In Figure In 2Figure, two di 2,ff twoerent different experiments, experiments, which yield which a and a rudimentary analysis of the particle distribution. In Figure 2, two different experiments, which similaryield a particlesimilar particle size from size TEM from and TEM XRD and analysis XRD areanalysis compared. are compared. While the While mean the diameter mean diameter is between is yield a similar particle size from TEM and XRD analysis are compared. While the mean diameter is 9between and 10 nm,9 and for 10 both nm, experiments, for both experiments, the standard the deviation standard isdeviation completely is completely different, which different, indicates which a between 9 and 10 nm, for both experiments, the standard deviation is completely different, which diinffdicateserent dispersity. a different dispersity. indicates a different dispersity.

FigureFigure 2.2. TEMTEM picturepicture of of sample sample 21 21 (a ()a and) and particle particle size size distribution distribution obtained obtained from from picture picture analysis analysis (b). Figure 2. TEM picture of sample 21 (a) and particle size distribution obtained from picture analysis TEM(b). TEM picture picture of sample of sample 2 (c) 2 and (c) particleand particle size distributionsize distribution obtained obtained from from picture picture analysis analysis (d). (d). (b). TEM picture of sample 2 (c) and particle size distribution obtained from picture analysis (d).

Crystals 2020, 10, 214 5 of 12

WhileCrystals the 2020, mean 10, x FOR diameter PEER REVIEW isbetween 9 and 10 nm, for both experiments, the standard5 of 12 deviation is completely different, which indicates a different dispersity. Thus, the synthesis parameters also influence theWhile particle the mean size diameter distribution is between and not9 and only 10 nm, the for primary both experiments, particle size. the Parametersstandard deviation of experiment 21 leadis tocompletely a small different, size distribution which indicates (σ = 0.891a different), while dispersity. experiment Thus, the 2 yields synthesis a broad parameters size distributionalso influence the particle size distribution and not only the primary particle size. Parameters of (σ = 2.572). In the case of those two experiments shown here, the longer titration time of iron chlorides experiment 21 lead to a small size distribution ( = 0.891), while experiment 2 yields a broad size to the basedistribution (experiment ( = 2.572). 2) leadsIn the tocase a broaderof those two size experiments distribution. shown In here, order the to longer facilitate titration a near time to of statistical size distribution,iron chlorides at to least the base 100 (experiment particles are 2) countedleads to a perbroader synthesis. size distribution. Even though In order this to approach facilitate a does not yield completelynear to statistical statistical size distribution, data on the at least particle 100 particles size distribution, are counted weper usesynthesis. it as anEven indicator. though this Weapproach observe does an not influence yield completely of synthesis statistical parameters data on the on theparticle particle size distribution, size and visualized we use it as our an findings indicator. in Figure3. The ratio of iron salts to sodium hydroxide is the most significant parameter here We observe an influence of synthesis parameters on the particle size and visualized our findings and a stoichiometricin Figure 3. The ratio ratio of yieldsiron salts larger to sodium particles hydroxide than is either the most an excesssignificant of parameter iron chlorides here and or a an excess of sodiumstoichiometric hydroxide. ratio This yields was larger expected particles and than is either in good an excess agreement of iron chlorides with previous or an excess studies of [17,18]. Furthermore,sodium we hydroxide. were able This to was observe expected an influence and is in goodof reaction agreement time with and previous temperature studies as [17,18] well as. stirring rate andFurthermore, dosing rate. we Reaction were able time to observe and dosing an influence rate onlyof react moderatelyion time and aff ect temperature the particle as well size. as stirring rate and dosing rate. Reaction time and dosing rate only moderately affect the particle size.

18 18

16 16

14 14

12 12

10 10 Size (nm) Size (nm) 8 8

6 6

1 1 1 0 1 0 0.75 0.75 0.5 0.25 0.5 0.25 0 0.5 0 0.5 0.5 0.5 -0.5 0.25 -0.5 0.25 C: Temperature 0.75 A: FeCl D: Stirring rate 0.75 A: FeCl B: NaOH B: NaOH -1 0 -1 0 1 1 (a) (b)

18 18

16 16

14 14

12 12

10 10

Size (nm) 8 Size (nm) 8

6 6

1 1 1 1 0 0 0.75 0.5 0.75 0.5 0.25 0.25 0 0.5 0 0.5 0.5 0.5 -0.5 0.25 F: Reaction time -0.5 0.25 A: FeCl E: Dosing rate 0.75 A: FeCl 0.75 B: NaOH B: NaOH -1 0 -1 0 1 1 (c) (d)

FigureFigure 3. The 3. The predicted predicted particle particle diameter diameter (size) (size) depending depending on temperature on temperature and iron and salt/sodium iron salt /sodium hydroxide concentration while stirring rate and reaction time is maximal and dosing rate is minimal hydroxide concentration while stirring rate and reaction time is maximal and dosing rate is minimal (a). The predicted particle diameter (size) depending on stirring rate and iron salt/sodium hydroxide (a). The predicted particle diameter (size) depending on stirring rate and iron salt/sodium hydroxide concentration while temperature and reaction time is maximal and dosing rate is minimal (b). The concentrationpredicted while particle temperature diameter (size) and depending reaction on time dosing is maximal rate and iron and salt/sodium dosing rate hydroxide is minimal (b). The predictedconcentration particle while temperature, diameter (size) stirring depending rate and reaction on dosing time is maximal rate and (c). iron The predicted salt/sodium particle hydroxide concentration while temperature, stirring rate and reaction time is maximal (c). The predicted particle diameter (size) depending on reaction time and iron salt/sodium hydroxide concentration while stirring rate and temperature is maximal and dosing rate is minimal (d). Crystals 2020, 10, x FOR PEER REVIEW 6 of 12

diameter (size) depending on reaction time and iron salt/sodium hydroxide concentration while Crystalsstirring2020, 10 rate, 214 and temperature is maximal and dosing rate is minimal (d). 6 of 12

Powder XRD yields the crystal structure and with analysis of crystal defects, even the diameter of nanoparticlesPowder XRD can yields be theestimated. crystal structureFurthermore, and with it is analysispossible of to crystal differentiate defects, evenbetween the diameterthe phases of nanoparticlesmagnetite and can maghemite be estimated. even Furthermore, though they ithave is possible a very similar to differentiate spinel and between distorted the phases spinal magnetitestructure, andrespectively. maghemite A even differentiation though they havebetween a very magnetite similar spinel and and maghemite distorted spinalon the structure, nanoscale respectively. is quite Achallenging, differentiation due betweento the crystal magnetite defects and. Amaghemite method proposed on thenanoscale by Kim et isal. quite can challenging,be used, where due the to 440 the crystaland the defects. 511 reflection A method is compared proposed bydue Kim to the et al. sli canghtly be greater used, where lattice the constant 440 and of the maghemite 511 reflection [31]. isAnother compared possibility due to to the distinguish slightly greater between lattice magnetite constant and of maghem maghemiteite is [Raman31]. Another spectroscopy possibility [29,32] to. distinguishWith this technique, between the magnetite distance and between maghemite Fe-O and is Raman thus a different spectroscopy vibration [29,32 constant]. With thisleading technique, to two thedistinct distance bands between in the spectrum Fe-O and can thus be aanalyzed. different Thi vibrations means constant that the leadingdistortion to of two the distinct spinel structure bands in theduring spectrum the oxidation can be process analyzed. leads This to two means Fe-O that distances the distortion which can of be the observed spinel structure around 660 during and 710 the 1 oxidationcm−1 in a Raman process spectrum. leads to two Schwaminger Fe-O distances et al. which have candemonstrated be observed an around analysis 660 of and these 710 bands cm− inand a Ramantheir ratios spectrum., which Schwaminger help to identify et al. havethe amount demonstrated of magnetite an analysis and of maghemite these bands in and an their iron ratios, oxide whichnanoparticle help to [28] identify. However, the amount multiple of effects magnetite due to and distortion maghemite and insurface an iron effects oxide cannot nanoparticle be identified [28]. However,with this method. multiple Furthermore, effects due to a distortion change of and the surfaceFe-O distance effects cannotdoes not be always identified correlate with this to a method. change Furthermore,in the crystal a structure, change of the differences Fe-O distance between does XRD not always and Raman correlate spectrum to a change of the in the same crystal particles structure, are diexpectedfferences [28] between. A com XRDparison and Raman of our spectrum phase analysis of the same is shown particles in Figure are expected 4 and [the28]. full A comparison data set is ofpresented our phase in analysisTable S4. is We shown were in not Figure able4 to and detect the fullany dataother set iron is presentedoxide or oxyhydroxide in Table S4. We phases were with not ableXRD to or detect Raman any sp otherectroscopy. iron oxide or oxyhydroxide phases with XRD or Raman spectroscopy.

100 26 XRD Magnetite (660) Raman Maghemite (710) 80 Cumulative

60

40 Intensity(n.u.) 25 Magnetite ratio (%) ratio Magnetite 20

0 200 400 600 800 1000 1200 0 10 20 30 40 Wave number (cm-1) Experiment number

(a) (b)

Figure 4.4. Raman spectraspectra ofof samplesample 25 25 and and 26 26 which which illustrate illustrate the the deconvolution deconvolution fitting fitting (a). (a) Correlation. Correlation of 1 magnetiteof magnetite ratios ratios obtained obtained from from Raman Raman spectroscopy spectroscopy (Ratio (Ratio between between 660 and 660 710 and cm 710− band)cm−1 band) and XRD and (440XRD reflection) (440 reflection) for all for samples all samples according according to their to numbertheir number in the in design the design of experiments of experiments (b). (b).

While KimKim etet al. have demonstrated that XRD is a useful tool to distinguish between maghemite and magnetite [[31]31],, wewe wantwant toto emphasizeemphasize thethe interpretationinterpretation ofof RamanRaman spectraspectra forfor ironiron oxideoxide phasephase analysis. A great advantage of the analysis of Raman spectra is that non-crystallinenon-crystalline and extremely smallsmall particlesparticles cancan be investigated. We investigated the influenceinfluence ofof the ratio between educts sodium hydroxide and iron chlorides on the phase composition as well as the influenceinfluence of temperature, stirring rate,rate, dosingdosing raterate andand reactionreactiontime time. (Figure A higher5). Amagnetite higher magnetite content can content be observed can be observed for high forsodium high hydroxide sodium hydroxide and low andiron lowchloride iron chlorideconcentrations. concentrations. The temperature The temperature and the stirring and the rate stirring only rateslightly only affect slightly the amagnetiteffect the magnetite content in content the iron in oxide the iron nanoparticles. oxide nanoparticles. Lower temperatures Lower temperatures and lower andstirring lower rates stirring lead to rates a hig leadher magnetite to a higher content magnetite than contenthigher temperatures than higher temperaturesand higher stirring and higher rates. stirringThe dosing rates. Therate dosing does ratenot does significantly not significantly influence influence the magnetite the magnetite content content of of the the synthesized nanoparticles. The reaction time influencesinfluences the content of magnetite.magnetite. Higher reaction times lead to a higher magnetite content.content.

Crystals 2020, 10, 214 7 of 12 Crystals 2020, 10, x FOR PEER REVIEW 7 of 12

100 100

80 80

60 60

40 40

20 20 Magnetite (%) Magnetite (%) 0 0

1 1 1 1 0 0 0.5 0.75 0.5 0.75 0.25 0.25 0 0.5 0 0.5 0.5 0.5 -0.5 0.25 C: Temperature 0.75 A: FeCl D: Stirring rate -0.5 0.25 A: FeCl B: NaOH 0.75 -1 0 -1 0 B: NaOH 1 1 (a) (b)

100 100

80 80

60 60

40 40

20 20 Magnetite (%) Magnetite(%) 0 0

1 1 1 1 0 0 0.5 0.75 0.5 0.75 0.25 0.25 0 0.5 0 0.5 0.5 0.5 -0.5 0.25 E: Dosing rate 0.75 A: FeCl F: Reaction time -0.5 0.25 A: FeCl B: NaOH 0.75 -1 0 -1 0 B: NaOH 1 1 (c) (d)

FigureFigure 5. 5. The predicted magnetite magnetite share share in in the the iron iron oxide oxide nanoparticles nanoparticles obtained obtained from from Raman Raman spectroscopyspectroscopy dependingdepending on temperature temperature and and iron iron salt/sodium salt/sodium hydroxide hydroxide concentration concentration while while stirring stirring raterate and and reaction reaction time time is maximal and and dosing dosing rate rate is is minimal minimal (a) (a. ). The The predicted predicted magnetite magnetite share share dependingdepending on on stirring stirring rate and and iron iron salt/sodium salt/sodium hydroxide hydroxide concentration concentration while while temperature temperature and and reactionreaction time time isis maximalmaximal and dosing dosing rate rate is is minimal minimal (b) (b. ).The The predicted predicted magnetite magnetite share share depending depending on on dosingdosing rate rate and and iron iron salt salt/sodium/sodium hydroxide hydroxide concentration concentration while while temperature, temperature, stirring stirring rate rateand reactionand reaction time is maximal (c). The predicted magnetite share and iron salt/sodium hydroxide time is maximal (c). The predicted magnetite share and iron salt/sodium hydroxide concentration while concentration while stirring rate and temperature is maximal and dosing rate is minimal (d). stirring rate and temperature is maximal and dosing rate is minimal (d).

AnAn evaluationevaluation of the model is is important important in in order order to to compare compare and and understand understand the theprediction prediction models for iron oxide nanoparticle synthesis. The correlation between the parameters and the models for iron oxide nanoparticle synthesis. The correlation between the parameters and the obtained obtained values can be described with the DoE model. Size and magnetite content prediction models values can be described with the DoE model. Size and magnetite content prediction models are based are based on linear mixture models. The predicted particle size values by the model are compared to on linear mixture models. The predicted particle size values by the model are compared to the actual the actual values obtained from TEM analysis (Figure 6a) The model predicts the particle size values obtained from TEM analysis (Figure6a) The model predicts the particle size significantly with a significantly with a p value of 0.0011. The higher p value compared to the study of Roth et al. (p < p0.0001) value ofcan 0.0011. be explained The higher by more p value parameters compared and to a thelarger study parameter of Roth range et al. [17] (p <. The0.0001) model can describing be explained bythe more magnetite parameters content and is a significantlarger parameter as well range with [17 a p]. Thevalue model of 0.0062 describing (Figure the 6b). magnetite However, content the is significantdistribution as of well the withcorrelation a p value between of 0.0062 actual (Figure and predicte6b). However,d values is the broader distribution than the of correlation the correlation of betweenthe size prediction actual and model. predicted values is broader than the correlation of the size prediction model.

Crystals 2020, 10, 214 8 of 12 Crystals 2020, 10, x FOR PEER REVIEW 8 of 12 Predicted vs. Actual Predicted vs. Actual

16 100

14 80

12 60 Predicted 10 Predicted 40

8 20

6 0

6 8 10 12 14 16 0 20 40 60 80 100 Actual Actual (a) (b)

FigureFigure 6. 6.PredictedPredicted pa particlerticle sizes sizes in incomparison comparison to tothe the actual actual particle particle sizes sizes (a) (a. ).Predicted Predicted magnetite magnetite contentcontent in incomparison comparison to tothe the actual actual magnetite magnetite content content (b ()b. ).

4. 4.Discussion Discussion TheThe study study focusses focusses on on two two aspects aspects of ofthe the co co-precipitation-precipitation of ofmagnetic magnetic iron iron oxide oxide nanoparticles. nanoparticles. OnOn the the one one hand, hand, how how can can we we control control and and influence influence the the size size of of iron iron oxide oxide nanoparticles? nanoparticles? On On the the other other hand,hand, how how is isthe the iron iron oxide oxide phase phase influenced influenced by by the the synthesis synthesis conditions? conditions? TheThe particle particle size size is issignificantly significantly affected affected by by the the ratio ratio of of iron iron chloride chloride to to sodium sodium hyd hydroxide.roxide. If Ifthe the ratioratio between between iron iron ions ions and and sodium sodium hydroxide hydroxide is in is the in therange range of a of 3:10 a 3:10 molar molar ratio ratio of iron of ironions ionsto sodiumto sodium hydroxide hydroxide (corresponding (corresponding to parameters to parameters of A = ofB = A 0.5),= B the= 0.5),co-precipitated the co-precipitated particles particleswill be largerwill bethan larger for other than conditions. for other conditions. This behavior This is behaviorexpected isand expected has been and demonstrated has been demonstrated by Roth et al. by andRoth Forge et al. et andal. [17,18] Forge. A et higher al. [17, 18amount]. A higher of sodium amount hydroxide of sodium will hydroxide lead to more will seeds lead since to more the seeds co- precipitationsince the co-precipitation starts in a supercritical starts in regime a supercritical which will regime then lead which to smaller will then crystals. lead to If smallerthe amou crystals.nt of sodiumIf the amounthydroxide of sodiumis too low, hydroxide seeding is will too take low, longer, seeding but will the take crystal longer, grow butth the will crystal stop growthat a point will wherestop atthe a pH point drops where below the the pH needed drops belowconditions the neededfor iron conditionsoxide co-precipitation. for iron oxide The co-precipitation. temperature playsThe temperature an important plays role an for important seeding role and for crystal seeding growth and crystal [16,33] growth. Especially [16,33 ]. the Especially Ostwald the ripening Ostwald processripening is process strongly is stronglydependent dependent on the ontemperature. the temperature. Thus, Thus, a higher a higher temperature temperature leads leads to to larger larger particles,particles, which which is isin in accordance accordance with with the the study study of of Forge Forge et et al. al. and and consisten consistentt with with the the nucleation nucleation modelmodel of ofLaMer LaMer and and Dinegar Dinegar [18,34] [18,34. The].The temperature temperature affects aff ectsthe solubility the solubility of iron of salts iron and salts the and pH the value.pH value. On the On other the hand, other a hand, higher a highertemperature temperature would wouldlead to leada higher to a nucleation higher nucleation rate than rate a lower than a temperaturelower temperature [33]. Thus [33, ].a higher Thus, atemperature higher temperature leads to leadsmore toseeds more followed seeds followed by a shorter by a growth shorter phase growth andphase therefore and therefore smaller smaller nanoparticles. nanoparticles. Roth etRoth al. etobserved al. observed the thesame same controversial controversial behavior behavior with with increasingincreasing temperatures temperatures [17] [17. Aside]. Aside from from the thetemperature, temperature, the stirring the stirring rate ratehas a has significant a significant effect e ffonect particleon particle size. size. We observe We observe the smallestthe smallest nanoparticles nanoparticles with with the the highest highest (1000 (1000 rpm) rpm) and and the the largest largest nananoparticlesnoparticles with with the the smallest stirring stirring speed speed (100 (100 rpm) rpm) investigated. investigated. A higher A higher stirring stirring speed speed is related is reltoated a higher to a higher energy energy dissipation dissipation in the in system, the system, which which is related is related to a fasterto a faster seeding seeding and thusand thus more more seeds seedscompared compared to slower to slower stirring stirring rates. rates. The The number number of seedsof seeds is usuallyis usually reciprocally reciprocally related related to theto the size sizeof nanoparticles.of nanoparticles The. The same same eff ecteffect was was observed observed by by Forge Forge et al.,et al., who who also also proposed proposed the the largest largest iron ironoxide oxide nanoparticles nanoparticles by usingby using a slow a slow stirring stirring speed speed and aand high a high temperature temperature [18]. Therefore,[18]. Therefore, our studies our studiesconcerning concerning the parameters the parameters affecting affecting the particle the size particle of a co-precipitation size of a co-precipitation process to synthesize process iron to synthesizeoxide nanoparticles iron oxide nanoparticles agrees very well agrees with ve literaturery well with and literature considers and more considers parameters more than parameters previous thanstudies previous [17,18 studies]. Concerning [17,18]. theConcerning magnetite the content magnetite of iron content oxide of nanoparticles, iron oxide nanoparticles, multiple studies multiple discuss studiesa dependence discuss a of dependence particle size of and particle magnetite size and content magnetite and a dependence content and of a oxidativedependen conditionsce of oxidative on the conditionsiron oxide on phase the [ iron14,29 oxide]. We used phase the [14,29] same. parameterWe used the range same and parameter the same experiments range and the to obtain same a experiments to obtain a model predicting the oxidation state of iron oxide nanoparticles, which was evaluated with Raman spectroscopy and XRD. The magnetite content derived from XRD and Raman

Crystals 2020, 10, 214 9 of 12 model predicting the oxidation state of iron oxide nanoparticles, which was evaluated with Raman spectroscopy and XRD. The magnetite content derived from XRD and Raman spectroscopy show similarities but there are also multiple deviations, which can be explained by the orthogonal analytical measurement methods. Which method gives us the more accurate result for the phase content? Here it is quite challenging to answer this question. We want to show with our, that there is no black and white, no magnetite or maghemite but multiple phase mixtures and transition states. The phase was not dependent on the size of the synthesized nanoparticles nor on the environment, but the most important parameter is the sodium hydroxide content. We tested the co-precipitation under oxidative environment (air/open vessel) and still found conditions which allow the synthesis of “pure” magnetite nanoparticles. Therefore, the oxidative environment does not influence the phase to same amount as the concentration of the base. The higher the ratio of base to iron salts, the higher the probability of obtaining pure magnetite nanoparticles. This finding can be explained due to the thermodynamic stability of the magnetite phase in alkaline environments. The parameters temperature and stirring rate, which are most relevant for seeding and crystal growth due to energy input to the reactor only slightly influence the phase of the generated nanoparticles. Interestingly, the reaction time is the most decisive factor aside from the stoichiometric ratio of educts. This is unexpected since the reaction was not performed in an inert atmosphere or in a closed vessel. However, this behavior can be explained by the high stability of the magnetite phase in alkaline environments. This means a phase transformation from maghemite to magnetite can take place during the nanoparticle formation in a co-precipitation synthesis [20]. Both DoE models are significant due to their low p values. On the other hand, even well-fitting models should be tested and verified. We identified the important parameters and use a large parameter range for this investigation. However, due to the higher amount of parameters compared to Forge et al. and Roth et al., our model predicting the particle size shows a higher p value, a higher lack of fit and a lower confidence level [17,18]. The model is in good agreement with literature data on iron oxide synthesis conditions and now spans a large range of parameters for the prediction of size and phase of iron oxide nanoparticles.

5. Conclusions Iron oxide nanoparticles have been synthesized and their synthesis parameters have been systematically investigated with design of experiments. The particle size was investigated with TEM and XRD, while the composition was analyzed with XRD and Raman spectroscopy. The focus of this work is on the influence of synthesis parameters towards the chemical composition and the particle size for a co-precipitation reaction of iron salts in alkaline environment. The nanoparticles synthesized with this method can be varied in the range of 5–16 nm within the used parameter range. The size of iron oxide nanoparticles can be influenced significantly by the stirring speed and reaction temperature. Furthermore, the stoichiometric ratio between iron salts and sodium hydroxide influence both size and chemical composition of the materials. We were not able to observe the formation of any other iron oxide or hydroxide species other than magnetite or maghemite with these synthesis parameters. Raman spectroscopy allows for a fast and simple distinction between iron oxide phases. Here, Raman spectroscopy has proven to be a well-suited technique to evaluate the magnetite content of iron oxide samples, which allows to dispense complex analysis techniques such as Mössbauer spectroscopy. The goal of this work was to investigate how iron oxide nanoparticles can be tailored towards their size and composition with a simple and low-cost co-precipitation method without using an inert atmosphere. The understanding of parameters helps to tailor and design magnetic nanoparticles for multiple uses and to better understand the crystallization, particle growth and oxidation of iron oxides at the nanoscale. Thus, these models are a useful tool for all people who want to synthesize iron oxide nanoparticles cost-effectively and in a way that is tailored to their application. Crystals 2020, 10, 214 10 of 12

Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4352/10/3/214/s1, Table S1. The parameter range according to the design of experiments with 6 parameters. Table S2. Absolute values of parameters implemented in the DoE model for all experiments. Table S3. XRD results for all samples obtained from 311 reflections and TEM diameters. Table S4. Ratio of deconvoluted A1g bands and magnetite share obtained from Raman and XRD (440) reflections. Author Contributions: Conceptualization, S.P.S. and S.B.; methodology, S.P.S. and C.S.; validation, S.P.S. and C.S.; formal analysis, C.S. and S.P.S.; investigation, C.S. and S.P.S.; resources, S.B.; data curation, C.S. and S.P.S.; writing—original draft preparation, S.P.S.; writing—review and editing, S.B. and S.P.S.; visualization, S.P.S.; supervision, S.B. and S.P.S.; project administration, S.B. and S.P.S.; funding acquisition, S.B. All authors have read and agreed to the published version of the manuscript. Funding: We appreciate support from the German Research Foundation (DFG) and the Technical University of Munich (TUM) in the framework of the Open-Access Publishing Program and by TUM International Graduate School of Science and Engineering (IGSSE). Acknowledgments: The authors would like to thank Carsten Peters for support with TEM imaging and Tom Nilges for the provision of the X-ray diffractometer. Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

References

1. Wu, W.; Changzhong, J.; Roy, V.A.L. Recent progress in magnetic iron oxide-semiconductor composite nanomaterials as promising photocatalysts. Nanoscale 2015, 7, 38–58. [CrossRef][PubMed] 2. Baumgartner, J.; Dey, A.; Bomans, P.H.H.; Le Coadou, C.; Fratzl, P.; Sommerdijk, N.A.J.M.; Faivre, D. Nucleation and growth of magnetite from solution. Nat. Mater. 2013, 12, 310–314. [CrossRef][PubMed] 3. Kolhatkar, A.G.; Jamison, A.C.; Litvinov, D.; Willson, R.C.; Lee, T.R. Tuning the magnetic properties of nanoparticles. Int. J. Mol. Sci. 2013, 14, 15977–16009. [CrossRef][PubMed] 4. Laurent, S.; Forge, D.; Port, M.; Roch, A.; Robic, C.; Vander Elst, L.; Muller, R.N. Magnetic iron oxide nanoparticles: Synthesis, stabilization, vectorization, physicochemical characterizations, and biological applications. Chem. Rev. 2008, 108, 2064–2110. [CrossRef][PubMed] 5. Lu, A.-H.; Salabas, E.L.; Schüth, F. Magnetic nanoparticles: Synthesis, protection, functionalization, and application. Angew. Chem. Int. Ed. Eng. 2007, 46, 1222–1244. [CrossRef] 6. Schwaminger, S.P.; Blank-Shim, S.A.; Scheifele, I.; Pipich, V.; Fraga-García, P.; Berensmeier, S. Design of Interactions Between Nanomaterials and Proteins: A Highly Affine Peptide Tag to Bare Iron Oxide Nanoparticles for Magnetic Protein Separation. Biotechnol. J. 2019, 14, e1800055. [CrossRef] 7. Schwaminger, S.P.; Fraga-García, P.; Blank-Shim, S.A.; Straub, T.; Haslbeck, M.; Muraca, F.; Dawson, K.A.; Berensmeier, S. Magnetic One-Step Purification of His-Tagged Protein by Bare Iron Oxide Nanoparticles. ACS Omega 2019, 4, 3790–3799. [CrossRef] 8. Roth, H.-C.; Schwaminger, S.P.;Peng, F.; Berensmeier, S. Immobilization of Cellulase on Magnetic Nanocarriers. ChemistryOpen 2016, 5, 183–187. [CrossRef] 9. Rossi, L.M.; Costa, N.J.S.; Silva, F.P.; Wojcieszak, R. Magnetic nanomaterials in catalysis: Advanced catalysts for magnetic separation and beyond. Green Chem. 2014, 16, 2906. [CrossRef] 10. Schnell, F.; Kube, M.; Berensmeier, S.; Schwaminger, S.P. Magnetic Recovery of Cellulase from Cellulose Substrates with Bare Iron Oxide Nanoparticles. ChemNanoMat 2019, 5, 422–426. [CrossRef] 11. Lartigue, L.; Wilhelm, C.; Servais, J.; Factor, C.; Dencausse, A.; Bacri, J.-C.; Luciani, N.; Gazeau, F. Nanomagnetic sensing of blood plasma protein interactions with iron oxide nanoparticles: Impact on macrophage uptake. ACS Nano 2012, 6, 2665–2678. [CrossRef][PubMed] 12. Owusu, K.A.; Qu, L.; Li, J.; Wang, Z.; Zhao, K.; Yang, C.; Hercule, K.M.; Lin, C.; Shi, C.; Wei, Q.; et al. Low-crystalline iron oxide hydroxide nanoparticle anode for high-performance supercapacitors. Nat. Commun. 2017, 8, 14264. [CrossRef][PubMed] 13. Tian, L.-L.; Zhang, M.-J.; Wu, C.; Wei, Y.; Zheng, J.-X.; Lin, L.-P.; Lu, J.; Amine, K.; Zhuang, Q.-C.; Pan, F.

γ-Fe2O3 Nanocrystalline Microspheres with Hybrid Behavior of Battery-Supercapacitor for Superior Lithium Storage. ACS Appl. Mater. Interfaces 2015, 7, 26284–26290. [CrossRef][PubMed] Crystals 2020, 10, 214 11 of 12

14. Baaziz, W.; Pichon, B.P.; Fleutot, S.; Liu, Y.; Lefevre, C.; Greneche, J.-M.; Toumi, M.; Mhiri, T.; Begin-Colin, S. Magnetic Iron Oxide Nanoparticles: Reproducible Tuning of the Size and Nanosized-Dependent Composition, Defects, and Spin Canting. J. Phys. Chem. C 2014, 118, 3795–3810. [CrossRef] 15. Carvalho, M.D.; Henriques, F.; Ferreira, L.P.; Godinho, M.; Cruz, M.M. Iron oxide nanoparticles: The Influence of synthesis method and size on composition and magnetic properties. J. Solid State Chem. 2013, 201, 144–152. [CrossRef] 16. Fang, M.; Ström, V.; Olsson, R.T.; Belova, L.; Rao, K.V. Particle size and magnetic properties dependence on growth temperature for rapid mixed co-precipitated magnetite nanoparticles. Nanotechnology 2012, 23, 145601. [CrossRef] 17. Roth, H.-C.; Schwaminger, S.P.; Schindler, M.; Wagner, F.E.; Berensmeier, S. Influencing factors in the CO-precipitation process of superparamagnetic iron oxide nano particles: A model based study. J. Magn. Magn. Mater. 2015, 377, 81–89. [CrossRef] 18. Forge, D.; Roch, A.; Laurent, S.; Tellez, H.; Gossuin, Y.; Renaux, F.; Vander Elst, L.; Muller, R.N. Optimization of the Synthesis of Superparamagnetic Contrast Agents by the Design of Experiments Method. J. Phys. Chem. C 2008, 112, 19178–19185. [CrossRef] 19. Massart, R. Preparation of aqueous magnetic liquids in alkaline and acidic media. IEEE Trans. Magn. 1981, 17, 1247–1248. [CrossRef] 20. Ahn, T.; Kim, J.H.; Yang, H.-M.; Lee, J.W.; Kim, J.-D. Formation Pathways of Magnetite Nanoparticles by Coprecipitation Method. J. Phys. Chem. C 2012, 116, 6069–6076. [CrossRef] 21. Daou, T.J.; Grenèche, J.M.; Pourroy, G.; Buathong, S.; Derory, A.; Ulhaq-Bouillet, C.; Donnio, B.; Guillon, D.; Begin-Colin, S. Coupling Agent Effect on Magnetic Properties of Functionalized Magnetite-Based Nanoparticles. Chem. Mater. 2008, 20, 5869–5875. [CrossRef] 22. Fan, T.; Pan, D.; Zhang, H. Study on Formation Mechanism by Monitoring the Morphology and Structure

Evolution of Nearly Monodispersed Fe3O4 Submicroparticles with Controlled Particle Sizes. Ind. Eng. Chem. Res. 2011, 50, 9009–9018. [CrossRef] 23. Kluge, S.; Deng, L.; Feroughi, O.; Schneider, F.; Poliak, M.; Fomin, A.; Tsionsky, V.; Cheskis, S.; Wlokas, I.; Rahinov, I.; et al. Initial reaction steps during flame synthesis of iron-oxide nanoparticles. CrystEngComm 2015, 17, 6930–6939. [CrossRef] 24. Ramimoghadam, D.; Bagheri, S.; Hamid, S.B.A. Progress in electrochemical synthesis of magnetic iron oxide nanoparticles. J. Magn. Magn. Mater. 2014, 368, 207–229. [CrossRef] 25. Yathindranath, V.; Rebbouh, L.; Moore, D.F.; Miller, D.W.; van Lierop, J.; Hegmann, T. A Versatile Method for the Reductive, One-Pot Synthesis of Bare, Hydrophilic and Hydrophobic Magnetite Nanoparticles. Adv. Funct. Mater. 2011, 21, 1457–1464. [CrossRef] 26. Cheng, C.; Xu, F.; Gu, H. Facile synthesis and morphology evolution of magnetic iron oxide nanoparticles in different polyol processes. New J. Chem. 2011, 35, 1072. [CrossRef] 27. Cornell, R.M.; Schwertmann, U. The Iron Oxides. Structure, Properties, Reactions, Occurrences, and Uses, 2nd ed.; Completely rev. and Extended Ed.; Wiley-VCH: Weinheim, Germany, 2003; ISBN 9783527302741. 28. Schwaminger, S.P.; Bauer, D.; Fraga-García, P.; Wagner, F.E.; Berensmeier, S. Oxidation of magnetite nanoparticles: Impact on surface and crystal properties. CrystEngComm 2017, 19, 246–255. [CrossRef] 29. Rebodos, R.L.; Vikesland, P.J. Effects of oxidation on the magnetization of nanoparticulate magnetite. Langmuir 2010, 26, 16745–16753. [CrossRef] 30. El Mendili, Y.; Grasset, F.; Randrianantoandro, N.; Nerambourg, N.; Greneche, J.-M.; Bardeau, J.-F. Improvement of Thermal Stability of Maghemite Nanoparticles Coated with Oleic Acid and Oleylamine Molecules: Investigations under Laser Irradiation. J. Phys. Chem. C 2015, 119, 10662–10668. [CrossRef] 31. Kim, W.; Suh, C.-Y.; Cho, S.-W.; Roh, K.-M.; Kwon, H.; Song, K.; Shon, I.-J. A new method for the identification and quantification of magnetite-maghemite mixture using conventional X-ray diffraction technique. Talanta 2012, 94, 348–352. [CrossRef] 32. Jubb, A.M.; Allen, H.C. Vibrational Spectroscopic Characterization of Hematite, Maghemite, and Magnetite Thin Films Produced by Vapor Deposition. ACS Appl. Mater. Interfaces 2010, 2, 2804–2812. [CrossRef] Crystals 2020, 10, 214 12 of 12

33. Vayssières, L.; Chanéac, C.; Tronc, E.; Jolivet, J.P. Size Tailoring of Magnetite Particles Formed by Aqueous Precipitation: An Example of Thermodynamic Stability of Nanometric Oxide Particles. J. Colloid Interface Sci. 1998, 205, 205–212. [CrossRef][PubMed] 34. LaMer, V.K.; Dinegar, R.H. Theory, Production and Mechanism of Formation of Monodispersed Hydrosols. J. Am. Chem. Soc. 1950, 72, 4847–4854. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).