Electrorheology of SI-ATRP-modified graphene oxide particles with poly(butyl methacrylate): effect of reduction and compatibility with silicone oil

Surface-initiated atom transfer radical polymerization (SI-ATRP) was used to modify graphene oxide (GO) particles with poly(butyl methacrylate) (PBMA) chains. This procedure facilitated the single-step fabrication of a hybrid material with tailored conductivity for the preparation of a suspension in silicone oil with enhanced sedimentation stability and improved electrorheological (ER) activity. PBMA was characterized using various techniques, such as gel permeation chromatography (GPC) and 1H NMR spectroscopy. Thermogravimetric analysis through on-line investigation of the Fourier transform infrared spectra, together with transmission electron microscopy, X-ray photoelectron microscopy, and atomic force microscopy, were successfully used to confirm GO surface modification. The ER performance was investigated using optical microscopy images and steady shear rheometry, and the mechanism of the internal chain-like structure formation was elucidated. The dielectric properties confirmed enhanced ER performance owing to an increase in relaxation strength to 1.36 and decrease in relaxation time to 5 × 10−3 s. The compatibility between GO and silicone oil was significantly influenced by covalently bonded PBMA polymer brushes on the GO surface, showing enhanced compatibility with silicone oil, which resulted in the considerably improved sedimentation stability. Furthermore, a controlled degree of reduction of the GO surface ensured that the suspension had improved ER properties.


Introduction
Electroresponsive systems, also known as Smart Systems, are groups of materials whose physical properties can be varied by applying an external electric eld. 1 Electrorheological (ER) suspensions are a type of electroresponsive system, 2 which essentially comprise semiconducting particles that can be induced with dipoles upon application of an electric eld and insulating uid. 3 Particles homogenously dispersed in the medium can create internal chain-like structures upon application of an external electric eld. Such structure formation results in a change in viscosity of several orders of magnitude. 4 This process is completely reversible, with the system reverting to its initial state aer the electric eld is switched off. This phenomenon has found various industrial applications, such as in damping, 5 haptic displays, 6 and invasive surgery. 7 In the last two decades, various materials, mainly based on inorganic particles (titanium oxide 8 or silica 9,10 ), conducting polymers (poly(aniline), 11 poly(pyrrole), 12 or poly(diphenyl amine), 13 ) and their core-shell analogues 14,15 and hybrids, 16 have been used as the dispersed phase in ER suspensions. Recently, Choi et al. 17 found that graphene oxide (GO) is a promising material for application in ER suspensions. Therefore, various researchers have extensively studied neat GO 18 and its core-shell particles 19,20 or hybrids, 21,22 and their effect on ER performance. Slight synergism has been obtained for GO core-shell particles, in which conducting polymers 23 or polar modifying agents play important roles, 24 owing to their tunable conductivity.
The surface of particles and nanoparticles can be modied by polymer chains through either reactions of polymer functional groups with the functional groups present on the particle surface 25,26 or direct polymerization, where "graing from" or "graing through" methods can be applied. 27 Various controlled/living polymerizations (CLP) allow "graing from" modications through surface-initiated polymerization, achieving polymer synthesis with controlled molar mass, narrow dispersity, and various functionalities. 28 Atom transfer radical polymerization (ATRP) is currently the most widely used reversible deactivation radical polymerization technique owing to the availability of transition metal catalysts and large-scale monomers, which can be polymerized using this technique. 29 Furthermore, ATRP initiator can easily be attached to various surfaces for subsequent surface-initiated ATRP (SI-ATRP). We recently reported that GO can be graed with polystyrene chains using SI-ATRP while simultaneously reducing the GO particles in the presence of tertiary amines commonly used as ligands in ATRP. 30 The GO reduction ability during SI-ATRP modication was an additional advantage of this ATRP technique. This approach was also conrmed to be valid for poly(glycidyl methacrylate), with the nal partially reduced and graed GO improving the ER performance of GO-based suspensions. 22 Owing to these promising results, other monomers, such as butyl methacrylate, were selected for GO graing in this study to further improve compatibility with silicon oil through the aliphatic chain present in the monomer structure, and because various inorganic/polymer hybrids based on materials with stronger dipole moments exhibit a synergistic effect on ER performance. [31][32][33][34] Experimental Materials Graphite (powder, <20 mm, synthetic) was used as a precursor for GO sheets. Sulfuric acid (H 2 SO 4 , reagent grade, 95-98%), sodium nitrate (NaNO 3 , ACS reagent, $99%), potassium permanganate (KMnO 4 , 97%), and hydrogen peroxide (H 2 O 2 , ACS reagent, 29.0-32.0 wt% H 2 O 2 basis) were used as chemical reagents to obtain suitable exfoliation conditions for forming GO sheets. a-Bromoisobutyryl bromide (BiBB, 98%) served as an initiator linked to the GO surface. Initiator bonding was performed in the presence of proton scavenger triethylamine (TEA, $99%). Butyl methacrylate (BMA, 99%), ethyl a-bromoisobutyrate (EBiB, 98%), N,N,N 0 ,N 00 ,N 00 -pentamethyldiethylenetriamine (PMDETA, $99%), copper bromide (CuBr, $99%), and anisole (99%) were used as a monomer, initiator, ligand, catalyst, and solvent, respectively. Diethyl ether (ACS reagent, anhydrous, $99%) was used as a drying agent. All chemicals were purchased from Sigma Aldrich (USA) and used without further purication (except BMA). BMA was puried by passing through a neutral alumina column to remove p-tert-butylcatechol as inhibitor prior to use. Hydrochloric acid (HCl, 35%, p.a.), ethanol (absolute anhydrous, p.a.), tetrahydrofuran (THF, p.a.), toluene (p.a.), and acetone (p.a.) were obtained from Penta Labs (Czech Republic). Deionized (DI) water was used in corresponding experimental processes and washing routines. Lukosiol M200 grade is silicone oil with a viscosity of 200 mPa s, as supplied by Lukosiol (Kolín, Czech Republic).
Preparation of GO modied with poly(butyl methacrylate) (GO-PBMA) GO particles were synthesized from graphite powder using a modied Hummers method, 35 as described in our previous study. 36 Hydroxyl groups on the GO surface were reacted with BiBB to covalently bond the ATRP initiator to the surface using the following procedure: GO (2 g), dried THF (60 mL), and TEA (12 mL) were mixed under an argon atmosphere at approximately 5 C (ice/water bath), and BiBB (7 mL) was added dropwise for 1 h, followed by further mixing at ambient temperature overnight. The reaction mixture was consecutively washed with THF (150 mL), acetone (150 mL), and nally water (2 Â 200 mL). Modied GO was then ltered off using a PTFE lter (pore size, 0.44 mm). Excess water was removed from the treated particles by washing with diethyl ether (3 Â 60 mL).
PBMA was graed from the GO surface using the SI-ATRP approach, as follows: GO containing bonded ATRP initiator (0.5 g) was placed into a Schlenk ask equipped with a gas inlet/ outlet and septum. The ask was evacuated and backlled with argon three times. The argon-purged chemicals, namely BMA (15.4 mL, 130.5 mmol), EBiB (0.192 mL, 1.305 mmol), PMDETA (1.090 mL, 5.220 mmol), and anisole (15 mL), were gradually added. The presence of oxygen was further minimized by degassing the system using several freeze-pump-thaw cycles. In a frozen state, the CuBr catalyst (187.2 mg, 1.305 mmol) was added under gentle argon ow and the ask was immersed into a silicone oil bath pre-heated to 60 C for 2 h to achieve polymerization. The molar ratio of the reactants (BMA/EBiB/CuBr/ PMDETA) was 100 : 1:1 : 4, while anisole (50 vol%) was used as solvent. During polymerization, the viscosity of the mixture gradually increased. The reaction was stopped by exposing the mixture to air and cooling to ambient temperature. The product was puried by ltration, washing with DMF (3 Â 100 mL) and diethyl ether (3 Â 50 mL). The nal product was dried in desiccator under ambient conditions.

Characterization techniques
The molar mass and dispersity of the PBMA chains grown from the sacricial initiator were determined by GPC using a PL-GPC220 instrument (Agilent, Japan) with THF as solvent at a ow rate of 1.0 mL min À1 . Polystyrene was used as a standard and anisole as an internal standard. Monomer conversions were determined by 1 H NMR using a 400 MHz VNMRS Varian NMR spectrometer equipped with a 5 mm 1H-19F/15N-31P PFG AutoX DB NB probe at 25 C and using deuterated chloroform as solvent. Infrared spectra of the llers were recorded on a Nicolet 6700 spectrometer (Thermo Scientic, USA) equipped with a SMART ATR accessory with Ge crystal. Raman spectra were measured on a Nicolet DXR spectrometer (Nicolet, USA) using an excitation wavelength of 532 nm (3 scans, resolution of 2 cm À1 ). The integration time was 30 s, while the laser power on the surface was set to 1 mW. Oxygen-containing groups present on the GO surface were investigated by thermogravimetric analysis (TGA/SDTA 851e, Mettler Toledo, Switzerland) using a heating rate of 10 K min À1 under a nitrogen atmosphere. Transmission electron microscopy (TEM) images were acquired with resolution of 0.35 nm using a Philips CM12 instrument (Philips, Amsterdam, Netherlands).
Conductivity was measured using the van der Pauw method (Keithley 6517B, USA) at room temperature. Powders of the synthesized neat GO and modied GO-PBMA samples were pressed into pellets (diameter, 13 mm; thickness, 0.3-0.4 mm) at 400 MPa. The density of the GO-based pellets was determined by weighing before and aer immersion in n-decane using a Sartorius R160P analytical balance (Sartorius AG, Germany).
XPS measurements were performed using a TFA XPS device from Physical Electronics. The base pressure in the XPS analysis chamber was approximately 6 Â 10 À8 Pa. The samples were excited over a spot area of 400 mm 2 using monochromatic Al Ka 1,2 radiation at 1486.6 eV. Photoelectrons were detected with a hemispherical analyser positioned at a 45 angle with respect to the surface normal of the sample. The energy resolution was approximately 0.5 eV. Survey-scan spectra were acquired at a pass energy of 187.85 eV, while for C1s, individual highresolution spectra were recorded at a pass energy of 29.35 eV and with an energy step of 0.125 eV. All spectra were referenced to the main C1s peak of the carbon atoms, which was assigned a value of 284.8 eV. The spectra were analysed using MultiPak v8.1c soware (Ulvac-Phi Inc., Kanagawa, Japan, 2006) from Physical Electronics (supplied with the spectrometer). C1s spectra were tted with a symmetrical Gauss-Lorentz function. A Shirley-type background subtraction was used.
Dried GO sheets on a mica surface were characterized using atomic force microscopy (AFM; Dimension Icon atomic force microscope, Bruker). Measurements were performed at a scan speed of 1 Hz with a resolution of 512 Â 512 pixels in ScanAsyst mode at room temperature under an air atmosphere. A ScanAsyst-Air probe with a resonant frequency of 70 kHz and a stiffness constant of 0.4 N m À1 (Bruker) was used. AFM data were processed using Gwyddion 2.51 soware (Czech Metrology Institute).
Suspensions (5 wt% of dispersed phase) were mixed according to the following procedure: for all measurements, GO particles or GO-PBMA analogues were dispersed in silicone oil by rst agitating mechanically for 5 min and then ultrasonicating for 30 s. Rheological measurements were performed in controlled shear rate (CSR) mode using a rotational viscometer (Bohlin Gemini, Malvern Instruments, UK). The suspensions were placed into parallel-plate geometry (diameter, 40 mm; gap, 0.5 mm). The electrorheological cell was connected to a high-voltage DC source (TREK 668B, USA) to generate electric eld strengths of 0-2.5 kV mm À1 . Before each measurement, the previously built-up particulate structures were destroyed by shearing the sample at a shear rate of 50 s À1 for 60 s. All measurements were performed at 25 C.
Suspensions containing 1 wt% of GO-based particles were mixed with silicone oil and injected between two copper electrodes (gap, 80 mm) and connected to a high-voltage source (Keithley 2410, USA). Development of the internal chain-like structures was recorded using an optical microscope (N 400M, China).
The ER mechanism of internal chain-like structure development was investigated using the power law model shown in eqn (1). 4 where s y is the yield stress, E is the electric eld strength, a is the particle response to electric eld application, and q is related to the internal structure stiffness. The dielectric properties were measured by impedance dielectric spectroscopy using a Novocontrol Concept 50 analyzer (Novocontrol, Germany) connected to cylindrical sample cell BDS 1307 for liquid materials. Dielectric properties, such as the relative permittivity (3 0 ) and dielectric loss factor (3 00 ), were investigated in the frequency range of 0.5 Hz to 2 MHz. Dielectric spectra were analysed using the Havriliak-Negami model (eqn (2)). 37 where D3 0 ¼ 3 0 s À 3 0 N is the dielectric relaxation strength, 3 0 s and 3 0 N are the relative permittivities at zero and innite frequencies (f), respectively, u is the angular frequency (2pf), t rel is the relaxation time, and a and b are shape parameters describing the asymmetry of the dielectric function.

Modication of GO with poly(butyl methacrylate)
The ATRP conditions for GO surface modication were set to obtain sufficiently long PBMA chains to improve the compatibility of the GO particles with silicone oil. This improved the stability of the nal suspension in a short polymerization time, during which slight reduction of the GO surface was also obtained to afford conductivity suitable for magnetorheological suspensions. Therefore, polymerization for 2 h, the BMA conversion was approximately 41%, while the molar mass and dispersity of the PBMA chains were 5920 g mol À1 and 1.23, respectively. The molar mass tted quite well with the theoretical mass and the dispersity was narrow, showing that ATRP of BMA under the applied conditions was well controlled. Furthermore, to carefully characterize the GO-PBMA hybrid particles, the graing density was calculated. There are various approaches to calculating the graing density, such as simple calculation from TGA, as used by Zhao et al. and Zhang et al. 38,39 However, in this case, a more precise calculation was used, involving TGA data, the molar mass of the polymer gras, and the shape parameters of the particles, as previously reported. 40,41 The GO particle density (r) of 2.54 g cm 3 , GO sheet thickness from AFM (l) of 2 nm, and calculated specic surface area (S) of 400 m 2 g À1 were successfully implemented. The graing density was then calculated to be 0.01 chain per nm 2 , showing a relatively low density of surface modication.
To conrm the successful modication of the GO particles with PBMA chains, the volatile products formed during TGA analysis were monitored on-line using FTIR. As shown in Fig. 1, the decomposition of oxygen-containing groups on the surface of GO was observed. The major portion of such substances were cleaved in the temperature range of 180-260 C (Fig. 1a), as reected in the C-OH absorption band at 1428 cm À1 , C]O band at 1723 cm À1 , and -OH resonance at 3510 cm À1 in the FTIR spectra (Fig. 1b). For sample GO-I, decomposition was observed in the temperature range of 150-250 C (Fig. 1c). This was indicated by both the original oxygen-containing groups and ATRP initiator moieties covalently attached to the GO surface, probably in close proximity, reecting the same absorption bands, but also to additional C-O-C vibrations at 1214 cm À1 and 1048 cm À1 and a C-H vibration at 2960 cm À1 from the initiator (Fig. 1d). For GO-PBMA, an additional peak appeared in the temperature range of 260-315 C that reected the decomposition of covalently bonded PBMA (Fig. 1e). The connection of this peak with the C]O absorption band at 1722 cm À1 and aliphatic chain C-H vibrations at 2953 cm À1 and 2893 cm À1 (Fig. 1f) conrmed the presence of PBMA on the GO sheet surface.
TEM images were used for visual conrmation of the PBMA layer on the GO surface. As clearly shown in Fig. 2a, neat GO was very well exfoliated and had only up to a few layers in its layered structure. In contrast, GO modication with PBMA was observed as a oss-like layer, which made the GO slightly darker and less sharp at the edges in the nal TEM image.
To support the TEM microscopy ndings, AFM images of both the neat GO and GO-PBMA particles were acquired (Fig. 2c  and d). These clearly showed that neat GO particles consisted of one or two layers (Fig. 2c) with 1 or 2 nm thick sheets along the particle width, respectively, similar to previously reported observations. 3,42 Darker parts of the gures represented additional particles, mostly at the nanoscale (several nanometres), or sheets perpendicular to the investigated sheet. In contrast, clear conrmation of the presence of a PBMA layer on the particle was observed when signicantly more individual GO-PBMA sheets were visible, due to improved dispersibility and signicant repulsion between the particles. Furthermore, particle thickness increased from 1 to 5 nm, showing that the polymer layer was around 3 nm thick, which corresponded well with the low graing density.
The surface properties of neat GO and GO-PBMA were characterized in detail using XPS. For GO surface modication, hydroxyl groups needed to be present on the GO surface to allow covalent bonding with the ATRP initiator for subsequent polymerization. XPS analysis (Table 3) showed that about 32% of carbon was bonded with oxygen through a single bond, which was expected to provide a sufficient amount of reactive sites for attachment of the ATRP initiator. Fig. 3 clearly shows that the amount of oxygen-containing groups decreased aer modication with the PBMA polymer shell, while the amount of carbon slightly increased. There were also some visible impurities, calculated as 0.5% in neat GO and 0.3% in GO-PBMA, respectively. GO-PBMA also showed a slightly visible bromine Br3d peak (Fig. 3b) that was part of the coating structure, showing a total content of 0.2%. Aer deconvolution, 95% of this area belonged to the covalently bonded C-Br (terminating the PBMA gra) as visible peaks at 72.0 eV and 70.5 eV, with only 5% showing free Br À (impurities from catalyst) as peaks at 68.9 eV and 67.6 eV. These results correlated well with those observed for brominated graphene. 43 Furthermore, the spectra intensities shown in Fig. 3b are lower due to substantial coating, similar to that reported by Li et al. 3 Further characterization of the GO and GO-PBMA surfaces was performed by C1s peak deconvolution, which identied individual oxygen-carbon groups, as summarized in Table 1. In accordance with the low graing density, only a slight increase in the C1s/O1s ratio was observed. Furthermore, aer SI-ATRP, an increase in the C1s sp 2 content was also observed resulting from partial reduction of the GO surface. GO reduction was also investigated using Raman spectroscopy and conductivity measurements, as presented below. Reduction of the GO surface during SI-ATRP can be expected to lead to partial cleavage of the graed polymer chains, providing a lower graing density, which is typical for surface- Paper initiated polymerizations. To prove this hypothesis, more detailed study is needed, which is outside the scope of the present study.

Reduction of GO during SI-ATRP
The reduction of GO particles plays an important role in their further application as a dispersed phase in ER suspensions because the values of electrical conductivity for neat GO are not sufficient for this type of application, while a slightly enhancement of conductivity by 2-3 orders of magnitude should provide a system with improved ER efficiency. 44 As mentioned above, the tertiary amine used as a ligand in ATRP can also act as a GO reducing agent. 30 Therefore, conductivity and Raman shi measurements of the GO pellets were performed to conrm the reduction of GO during the SI-ATRP of BMA. Although the conductivity of neat GO was approximately of 1 Â 10 À8 S cm À1 , the conductivity of GO aer the SI-ATRP of BMA was increased to 6 Â 10 À7 S cm À1 . The reduction of GO was also conrmed by calculating the peak intensities of D to G, which reected the sp 2 and sp 3 hybridized forms of GO, using Raman spectroscopy (Fig. 4). This showed that the recoverability of the conducting pathways on the GO surface was similar to those reported elsewhere. 30,45 I D /I G was calculated to change from 0.90 for neat GO to 1.09 for GO-PBMA. Furthermore, the 2D structure sustained aer modication indicated that retained layer was very thin, in good agreement with the TEM and AFM observations. Therefore, the density of the GO nanoparticles also changed only slightly from 2.54 g cm À3 for neat GO to 2.21 g cm À3 for GO-PBMA. These results clearly showed that the origin of electrical conductivity enhancement was based on GO reduction rather than other phenomena, such as electronic interactions, because even the polymer layer showed pronounced repulsion between individual sheet-like GO particles.

Optical microscopy of chain-like internal structures
Optical microscopy is a useful tool for investigating chain-like structures formed in the presence of an external electric eld. As shown in Fig. 5a, oxygen-containing groups present on the surface of neat GO particles could not ensure their sufficient dispersion in the silicone oil, resulting in marked GO aggregation. Furthermore, owing to low conductivity, the system based on neat GO was not able to develop proper internal structures aer applying an electric eld (Fig. 5b), resulting in only weak ER performance, as discussed in the next section. In contrast, GO-PBMA particles were well dispersed in the silicone oil and did not form aggregates owing to the substantial poly(butyl methacrylate) layer on the surface of the GO particles (Fig. 5c). Furthermore, such particles enabled a fast response to application of an electric eld and created well-developed chain-like structures (Fig. 5d) essential for good ER performance.

Steady shear investigations under an external electric eld
Steady shear rheological investigation is commonly used to quantify the ER capability of graphene oxide-based silicone oil suspensions. 18,20,31 As shown in Fig. 6a, the neat GO-based suspension exhibited near-Newtonian behaviour in the absence of an external electric eld. When an electric eld of 0.5 kV mm À1 was switched on, the formation of partial chainlike structures was clearly visible (Fig. 5b) and the behaviour became pseudoplastic, as reected by a yield stress of 4.6 Pa. Further increasing the external eld to 1.5 kV mm À1 and 2.5 kV mm À1 resulted in increased yield stresses of 24 Pa and 41 Pa, respectively. In contrast, GO modied with PBMA had increased particle conductivity and better interactions with    This journal is © The Royal Society of Chemistry 2019 PDMS, resulting in the considerably different behaviour of this sample. The behaviour was also near-Newtonian in the absence of an external eld, with a slight deviation resulting from the better interactions of GO-PBMA particles with silicone oil, similar to those already reported for GO-PGMA analogues. 46 Aer application of the external electric eld (0.5 kV mm À1 ) and formation of relatively strong chain-like structures (Fig. 5d), a yield stress of 10.5 Pa was achieved (Fig. 6b). Similar to previous reports, the yield stress further increased with increasing external electric eld, reaching nearly 110 Pa at 2.5 kV mm À1 , which was signicantly higher than the value obtained for a common suspension based on GO, of approximately 35 Pa. 17 Additional studies dealing with variously modied graphenes, GO, and newly reported materials and their ER performance are summarized in Table 2. Therefore, the results obtained from the GO-PBMA-based system were very promising because it contained only 5 wt% of particles in the suspension.
To investigate the mechanisms of internal chain-like structure formation, the dependence of yield stress on the electric eld strength was plotted and parameters of the power-law model t (Table 3) were used to investigate this phenomenon, similar to previous reports. 51,52 There are two mechanisms of internal structure formation. The rst is the conductivity mechanism, when the coefficient a from eqn (1) reaches 1.5, which is based on the conductivity mismatch between the particles and silicone oil. The second mechanism is the polarization mechanism, when coefficient a reaches 2, which reects the relative permittivity mismatch between semiconducting particles and silicone oil. 11,53,54 For suspensions of neat GO, the value of a was found to be 1.36 (Fig. 7), indicating the conductivity mechanism. However, the value was considerably deviated from 1.5, indicating structures that were not precisely developed, in good agreement with the optical microscopy results (Fig. 5b). For the GO-PBMA-based suspension, parameter a was determined to be 1.48, which was in good agreement with conductivity mechanism and the well-developed internal chain-like structures (Fig. 5d). Furthermore, the q parameter, reecting the rigidity of the chain-like structures, was found to be 11.2 and 27.7 for neat GO and GO-PBMA-based suspensions, respectively, which correlated well with results obtained from dielectric investigations described in a later section.

Dielectric properties
Dielectric properties are important characteristics for ER suspensions because there is a correlation between the ER performance and dielectric properties. 55 Two crucial parameters, namely, dielectric relaxation strength (D3 0 ) and relaxation    (2)), were shown to signicantly inuence the magnitude of ER performance. Generally, the ER performance increases with increasing D3 0 and decreasing t rel . Therefore, optimal dielectric properties were crucial for an ER system with enhanced ER effect. These values from neat GO-based suspensions were not appropriate, with D3 0 and t rel values of 0.59 and 0.21 s, respectively ( Fig. 8 and Table 4). In contrast, when GO was modied with PBMA, partial reduction of the GO and improved wettability with silicone oil (Fig. 8, inset image) resulted in improved dielectric characteristics, with D3 0 and t rel values of 1.37 and 5 Â 10 À3 s, respectively ( Fig. 8 and Table 4). Such signicant improvements were well-correlated with the results obtained from both optical microscopy and electrorheological studies in the presence of an external eld, con-rming the enhanced ER performance of the GO-PBMA-based suspensions. Therefore, this system seems highly promising in comparison with other GO or layered systems, as summarized in Table 2.

Compatibility with silicone oil
The nal part of this study focused on the compatibility of neat GO and GO-PBMA particles with silicone oil. This behaviour was crucial for the real applicability of these systems. Very weak compatibility of the dispersed phase with silicone oil would be expected to lead to signicant particle sedimentation, because neat GO has a 2.5-times higher density than neat silicone oil and neat GO has poor wettability. Therefore, three experiments to determine the magnitude of compatibility between the particles and liquid medium were performed, namely, contact angle measurements, rheological investigation of the shear viscosity, and a sedimentation test. The neat GO particles showed a relatively high contact angle of 47.8 AE 2.9 (Fig. 9b, inset image (A)) due to rather small interactions of neat GO with the dispersed phase, and constant shear viscosity values at various shear rates (Fig. 9b). Finally, neat GO showed very poor sedimentation stability (Fig. 9a). In contrast, GO modied with PBMA chains showed signicantly improved wettability with silicone oil. The contact angle decreased to 26.8 AE 2.3 (Fig. 9b, inset image (B)), reecting enhanced interactions of the GO-PBMA particles with silicone oil and resulting in more pseudoplastic behaviour being observed (Fig. 9b). Consequently, improved sedimentation stability was also observed for GO-PBMA (Fig. 9a), with a sedimentation ratio three times higher than that of neat GO, while the particle density had only decreased to 2.21 g cm À3 . Therefore, it could be stated that the improved sedimentation stability was mostly caused by improved compatibility, while a lower density probably only marginally contributed to this decrease.

Conclusion
In this study, the modication of GO particles with PBMA chains using a SI-ATRP approach was performed to conrm that this technique was promising for the development of novel dispersed phases for ER systems. The polymerization of BMA was found to be well controlled regarding molar mass and dispersity. The compact coating of GO with PBMA was conrmed by TGA-FTIR, TEM, and AFM investigations. Targeted partial and simultaneous reduction of GO during SI-ATRP was conrmed by XPS analysis, the conductivity increased from 1 Â 10 À8 S cm À1 to 6 Â 10 À7 S cm À1 , and the I D /I G ratio increased from 0.90 to 1.09 in the Raman spectra. These ndings were promising for applications in ER uids, which were elucidated in detail using optical microscopy, steady shear rheology, and dielectric property measurements. Suspensions based on the GO-PBMA hybrid system exhibited well-developed internal chain-like structures with a yield stress of 110 Pa, dielectric relaxation strength of 1.37, and relaxation time of 5 Â 10 À3 s. Finally, the real-life applicability of this system was conrmed by the signicantly enhanced sedimentation stability caused by enhanced interactions with silicone oil, as conrmed by the decrease in contact angle and shear viscosity prole. Therefore, the modication of GO with PBMA chains accompanied by partial GO reduction provided an ER system with enhanced sedimentation stability and improved ER properties compared with the system based on neat GO or previously described systems based on GO modied with acrylate-based polymers.

Conflicts of interest
There are no conicts to declare.