The use of lanthanides as probe ions and signal transducers in sensing applications is widely accepted, as they can often provide sensitivity in the parts per trillion range and lower [1]. Lanthanides comprise the largest naturally occurring group in the periodic table, and their similarities arise from a resemblance in the electronic configurations of the elements which consists of the xenon levels, filled 6s sublevel, and a varying amount of electrons occupying the 4f sublevel. Generally, the optical absorption and emission spectra of the ions formed by the triply charged free lanthanide ions consist of very narrow lines (0.1 nm-0.01 nm). Quantum mechanical calculations have indicated that the energies and the radial extensions of the 4f eigenfunctions dramatically drop at the beginning of the lanthanide series so that the maxima of the 4f eigenfunctions no longer exceed those of the 5s and 5p eigenfunctions [2]. Thus, the 4f orbitals are not the outermost shell, but are largely shielded from the external environment by the filled 5s and 5p shells. Since these 4f orbitals are shielded, the electrostatic field in lanthanide complexes, induced by coordinating ligands causes only small perturbations in the 4f electrons energy levels. As a result, the lanthanides within the complex retain their semi “atomic” nature which gives the characteristic narrow bands observed in lanthanide emission and excitation spectra [3,4].

These narrow features often result in analyses which are both selective and sensitive. The spectral position and intensity of these features are dependent on the environmental symmetry of the lanthanide complex [5]. As a result of this unique chemistry, lanthanides have been used for a variety of purposes such as catalysts, optical components, alternatives to radioisotopes, and as scientific probes for a wide variety of phenomena including probes of temperature [6,7] and structure (both electronic and crystalline) [8,9]. Trace analysis techniques based on lanthanide spectroscopy have also found application for detection of many compounds often at parts per trillion (ppt) levels or lower [10-12].

Europium is the most commonly used lanthanide for sensing applications for several reasons. First, the spectral splitting patterns of europium are generally less complex (fewer peaks) and easier to interpret than the other lanthanides. Additionally, europium has the ⁵D₀→⁷F₂ transition which is a “hypersensitive transition” reflecting even minor changes in its environment with very intense luminescence [13] (In this case the term luminescence is preferred over other terms such as fluorescence or phosphorescence because the intra-configurational transitions may or may not involve a change in spin) [5]. Europium is also the most reactive of the lanthanides because unlike most lathanides which are primarily trivalent, europium also exists in the +2 state (electron configuration ₄F⁷) and the half-filled f-shell gives more stability [14].

Complexation by organic ligands was found to significantly enhance the luminescence intensity of the tripositive lanthanide, Ln(III) ions [1,14-17]. This enhancement was explained by a ligand to metal energy transfer mechanism [18,19]. The proposed mechanism indicates that when an excited triplet state of a coordinating ligand overlaps an excited lanthanide electronic level, the lanthanide luminescence is effectively pumped by the large cross section molecular absorbance of the ligand, rather than by the weak lanthanide absorbance. This process is believed to be much more efficient than direct absorption of light by the lanthanide since the lanthanides themselves exhibit low molar absorptivities. Proper ligand choice is therefore crucial in the preparation of lanthanide complexes.

Graphene oxide was selected as a coordinating ligand for the lanthanides because of its unique structural, electronic and optical properties [20], as well as its high mechanical strength [21]. The complexation with graphene oxide would allow the lanthanide to be incorporated into a wide range of materials including field emission displays [22-24], sensors [25-27], thin film transistors [28-30], touch panels [31], electrodes for optoelectronic devices [32,33], fuel cells [34-36], solar cells [37,38], and batteries [39,40]. Graphene oxide was chosen over graphene because the intrinsic oxygen was needed to coordinate with the lanthanide. Limited studies have addressed the luminescence properties of graphene and graphene oxide [31]. However, the success has been limited by a lack of uniformity in treatment techniques and the inability to make materials that are strongly luminescent. The ability of graphene to quench the luminescence of fluorescent dyes has also been reported [41,42]. Lanthanides, on the other hand, have unique and strong luminescence properties. Combining the luminescence properties of lanthanides with the structural properties of graphene oxide was a natural step towards the creation of luminescent graphene materials.

In this study, europium, terbium, gadolinium, samarium or dysprosium were selected as potential candidates, since they have partially filled subshells which provide them with enhanced sensitivity to the environment and increased luminescence properties. Since each lanthanide has luminescence in a different region of the spectra, incorporating several provided graphene oxide materials that luminesced in many colors. The resulting europium, terbium, samarium, dysprosium, and gadolinium complexes with graphene oxide (GO), were verified by raman, and laser induced luminescence. The surface morphology and physical properties of the graphene/ lanthanide materials (europium and terbium) were investigated by SEM and TEM.

Reagents

Unless otherwise indicated, materials were obtained from commercial suppliers and used without further purification. Analytical reagent grade chemicals were used along with deionized water to prepare solutions. The lanthanides were obtained from Aldrich (Aldrich, Milwaukee, WI). The graphene oxide was prepared by Teresa Bandosz at the City College of New York (CUNY) and used as received [43]. Eu(NO₃)₃. 5H₂O was prepared by taking Eu₂O₃ and dissolving it in water with heat and just enough nitric acid to produce a clear solution. The resulting solution was evaporated to near dryness, and then left to crystallize. The resulting solid was evaluated spectroscopically to confirm the conversion. The remaining lanthanides were purchased as nitrates and used as obtained.

Instrumentation

Luminescence was measured using a model 35-LAP-321-120 argon ion laser for excitation (Melles Griot, Carlsbad, CA). Holographic filters at 488 nm and 514 nm (Kaiser Optical Systems, Ann Arbor, MI, USA) were used to exclude all other laser lines. Spectra were collected using an f/4, 0.5-m monochromator (Chromex, Albuquerque, NM) equipped with a model ST-6 CCD (Santa Barbara Instruments Group, Santa Barbara, CA) using Kestrel Spec Software (K&M Co., Torrance, CA). Spectra were also obtained with a StellarNet Blue wave miniature fiber optic spectrometer (StellarNet, Tampa, FL) with a range 500-700 nm equipped with a 2400 g/mm ruled grating, a 2048 element detector array, integrated order sorting filter, 16-bit digitizer, and a 25 µm slit and running SpectraWiz software. The excitation and emission light was carried to and from the sample using a 400 µm bifurcated multimode fiber (Thor Labs, Newton, NJ). Scanning electron micrographs were obtained using a JEOL 6300F Field Emission Scanning Electron Microscope SEM (JEOL, Ltd., Tokyo).

Transmission Electron Micrographs were obtained by placing small amount of powder into a vial of Methanol and ultrasonicating them using a Branson B-12 80 watt ultrasonic cleaner for 5 minutes. Once the particles were suspended, a small amount of liquid was drawn off using a pipette and placed drop-wise onto a holey carbon film copper TEM grid purchased through SPI Supplies and allowed to air dry on clean filter paper. Prior to use, the holey carbon film TEM grids were pre-cleaned with de-ionized water to remove as much adsorbed silicon and other detritus from SPI Supplies manufacturing process. The particle samples were then imaged using a JEOL FasTEM 2010 TEM at 200 kv using a Gatan Corp. Gatan Image Filter (GIF 2000) digital camera and the Energy Dispersive X-ray Spectrometry was recorded using an EDAX Corp. Genesis Si(Li) 30 mm² EDS detector system with the TEM operated in S (TEM) mode.

Raman samples were analyzed on an aluminum substrate using a Jasco NRS-3000 (Jasco Inc. Easton, MD) at 229 nm, 532 nm, and 785 nm. Samples were analyzed using a 2400 groove/mm grating with slits set to 100 microns and integration times between 20-30 seconds depending on the sample.

Compound preparation

Graphene oxide was prepared by Teresa Bandosz, (City College New York) and used as obtained. Graphene oxide/europium materials were first prepared following the procedure outlined by Kim et al. [44]. The graphene oxide (0.5 g), water (20 mL) and NaOH (1 M aqueous solution, 1 mL) were sonicated for 30 minutes in a Cole-Parmer sonicator at 200 W. Europium nitrate, Eu(NO₃)₃. 5H₂O (0.428 g, 1 mmol) was added and stirred overnight. The solids were centrifuged, washed with water and acetone. The graphene/europium (Eu(GO)) solids were air dried on a watch glass overnight. This procedure was repeated with increasing amounts of europium nitrate (1.0 g, 1.25 g, 1.5 g and 2.0 g). The same procedure described above was repeated for terbium, gadolinium, samarium, and dysprosium nitrate.

Compound evaluation

The graphene oxide, europium oxide, lanthanide nitrates and the resulting lanthanide graphene oxide samples were evaluated using the laser induced fluorescence system described above. Excitation wavelengths of 465.8 nm, 488 nm, 496 nm, and 514 nm were evaluated for all materials. Optimal wavelengths were selected for each material: 488 nm for the europium, and samarium complexes, 254 nm for gadolinium and 514 nm selected for the terbium complexes. An optimal wavelength for dysprosium was not able to be determined using the excitation sources available.

The lanthanide graphene oxide compounds with the highest overall luminescence intensity, (europium and terbium graphene oxide) were selected for SEM and TEM.

As expected, the europium samples exhibited the greatest intensity (visibly red/orange) and interpretation of its spectral line patterns was easiest (based on number of peaks). Europium benefits from being 1 electron short of a half full shell making it easy for it to accept the electron for strong energy transitions. The conversion from europium oxide to europium nitrate was confirmed spectroscopically by a comparison with the spectra published by Buenzli and Yersin [45]. Europium samples were evaluated using excitation wavelengths of 465.8 nm, 472.7 nm, 488.0 nm, and 514.5 nm. Excitation at 465.8 nm was determined to have the highest luminescence intensity and give the best peak resolution, confirming previously published results [5]. Complexation between the lanthanide and the graphene oxide was determined by spectral changes and increased luminescence when compared with the luminescence of the lanthanide and the graphene oxide by themselves (Figure 1). The europium graphene oxide sample with the highest luminescent intensity was that prepared with 1.5 g of europium nitrate. The sample prepared with 0.5 g showed no spectral difference from the graphene oxide sample. This trend of highest luminescence for the complexes prepared 1.5 g of lanthanide nitrate was followed for all the lanthanides examined in this study, and it was these samples that were chosen for full characterization.

Figure 1: Spectra of the europium, graphene oxide, and the Eu(GO) showing the changes that occur with co-ordination. The europium and graphene oxide are shown on the right axis for better comparison. Excitation wavelength 488 nm.

In comparing the luminescence spectra of graphene oxide, europium nitrate and europium graphene oxide several transitions were evaluated. The ⁵D₀→ ⁷F₁ transition of the europium (from about 590-595 nm) exhibited a 1 nm shift when converted from the nitrate complex at 592 nm and the Eu(GO) complex at 591 nm. Additionally, the two peaks in the 595 nm region are better resolved in the Eu(GO) complex. The Eu(GO) peak at 595 nm is also higher in intensity, however this could be due to the fact that the GO itself has a fluorescence peak at that wavelength.

The hypersensitive ⁵D₀→ ⁷F₂ europium transition was also evaluated for evidence of Eu(GO) complex formation. The europium nitrate peak at 615 nm was shifted one nanometer to the blue (614 nm) in the Eu(GO) and also reduced in intensity relative to the peak at 617.5 nm. The peak at 617.5 nm was shifted 0.1 nm to the blue in the nitrate complex. Comparing these splitting patterns with the calculations done by Carlos et al. [15] (and taking into account no peaks were seen in the ⁵D₀→ ⁷F₁ transition) the site symmetry for both molecules is likely D₂ or D₂d with either a tetragonal or rhombic crystal system. If only one transition could be analyzed for binding, this would be the one chosen due to its sensitivity and intensity.

The terbium, samarium, gadolinium, and dysprosium samples were evaluated in a similar manner with the results summarized in Table 1. The terbium complexes (excited at 514 nm) exhibited the second strongest luminescence (visibly green) and had the next easiest spectra to evaluate (Figure 2). Terbium has higher intensity bands than expected because, like europium, it is one electron away from having a half full (stable) f shell. This makes promotion of the extra electron easier and gives rise to stronger luminescence. The first transition evaluated, the ⁵D₄→ ⁷F₅ had the strongest luminescence. In this spectral region, three peaks were seen for the terbium nitrate (541.5 nm, 543.3 nm, and 545.8 nm) while only two peaks were noted for the Tb(GO) complex (544.5 nm and 547.0 nm). The intensity of the largest peak also red shifted from the nitrate to the Tb(GO). In the ⁵D₄→ ⁷F₄ transition, the terbium nitrate has 2 unresolved peaks (581.0 nm and 585.25 nm) while the Tb(GO) complex has only one (585.0 nm). The ⁵D₄→ ⁷F₃ transition has one peak at 621.5 nm for the nitrate and two unresolved peaks at 621.3 and 622.8 nm in the Tb(GO) complex. The GO and Tb(GO) complex both had weak peaks at 536.8 nm and 593.8 nm which were not seen in the terbium nitrate spectrum.

Table 1: Major peaks in Graphene Oxide/Lanthanide spectra.

Figure 2: Spectra of the terbium, graphene oxide, and the Tb(GO) complex showing the changes that occur with co-ordination. The terbium nitrate and graphene oxides are shown on the right axis for better comparison. Excitation wavelength 514.0 nm.

The gadolinium, situated on the periodic table between europium and terbium does not have the same intense luminescence characteristics as its neighbors because it has the stability of a half filled f shell making it very difficult to promote an electron to the excited state. An excitation wavelength of 488 nm was used for the gadolinium complexes. 488 nm is not necessarily the optimal excitation wavelength for gadolinium but it was the strongest band output by the laser available, and gave the best results. UV excitation using a hand held type uv-light at 254 nm and 365 nm was also attempted with no success. As seen in the previous complexes the gadolinium graphene oxide had an overall luminescence increase over the nitrate and was much more intensly luminescent than the graphene oxide alone (Figure 3). New stong luminescence bands were seen in the 600 nm-620 nm region, as well as peak shifts and splitting changes in the 660 nm-670 nm, 740 nm-755 nm, and 785 nm- 800 nm regions. These changes were attributed to the complexation of the gadolinium and the graphene oxide. There are so many bands associated with gadolinium complexes that assigning the transitions with certainty was not possible, and spectral changes were indentified by region of the spectrum.

Figure 3: Spectra of gadolinium, graphene oxide, and the Gd(GO) complex excited at 254 nm.

Samarium, the lanthanide immediately preceeding europium has many absorption bands in the region below 500 nm and efficient fluorescence occurs from the ⁴G_5/2 level for solid materials. The emission is divided between many terminal levels making spectral interpretation difficult, and many ion-ion self quenching transitions are possible [13]. The preparation of the samarium complexes was also more difficult. The materials had to be purified by dissolving the materials in equal parts water/methanol and recrystallizing. Excited at 488 nm, the overall luminescence increase from the samarium nitrate to the samarium graphene oxide complex was not as intense as that seen for europium and terbium (Figure 4). Wavelengths of 254 nm, 465.8 nm, 472 nm, and 514 nm were also evaluated but gave no notable luminescence. As expected, the spectrum was much more complicated than that of terbium or europium making it less useful as a probe ion but still providing another color region for optical applications. Confirmation of complexation between the samarium and the graphene oxide was verified by spectral changes in several regions. In the transitions ⁴G_5/2→⁶H_5/2 (580 nm-600 nm region), ⁴G_5/2→ ⁶H_7/2 (600 nm-620 nm region), and ⁴G_5/2→ ⁶H_9/2 (640 nm-660 nm region) peak splitting changes were noted as well as the appearance as new intense luminescence bands. In the ⁴G_5/2→ ⁶H_11/2 transition (740 nm-770 nm region) splitting changes and peak shifts were noticed as well as an intensity increase. In the ⁴G_5/2→ ⁶F_7/2 transition (790 nm-820 nm region) splitting changes and peak shifts were observed between the nitrate and samarium graphene oxide complex. All of these indicate successful coordination of the materials.

Figure 4: Spectra of samarium, graphene oxide, and the Sm(GO) complex excited at 488 nm.

Dysprosium, the lanthanide immediately following terbium, was also reacted with graphene oxide and evaluated. Regardless of the exciation used, no changes were detected between the dysprosium nitrate and the dysprosium graphene oxide material. The synthesis was attempted many times using many different ratios with little luminescence reponse from any of them. The energy levels for dysprosium are typically very high making it hard to excite without complexation with organic ligands to provide intersystem crossing and it is surmised that the graphene oxide does not provide the levels needed. The raman spectra did exhibit significant changes between the nitrate and the graphene oxide complexes which supports this conclusion.

The SEM analysis was performed on the graphene oxide as well as the europium and terbium graphene oxide materials in an effort to better understand what was going on at the surface. The SEM of the graphene oxide showed the expected sheet like structure. In the europium graphene oxide as well as the terbium graphene oxide SEM images the sheet like structure can still be identified and the large lanthanide ions are seen to cover the surface (Figure 5).

Figure 5: SEM images of (rt) europium graphene oxide, (ctr) graphene oxide and (l) terbium graphene oxide.

TEM images were also collected for the europium and terbium graphene oxide particles (Figure 6). High resolution analysis of the europium graphene oxide particles shows limited graphene crystallinity. EDS data collected on the sample found the expected europium, oxygen, and carbon. The X-ray mapping illustrated the strong correlation of the europium being bound to the graphene oxide particles. The lower magnification TEM images for the terbium graphene oxide particles sample showed the graphene present as mostly planar flakes. The high magnification TEM indicated that the particles exhibited good crystallinity of the graphene as the atomic lattice features of the graphene particles were clearly visible. The EDS spectra confirmed the prescence of terbium, oxygen, and carbon in the sample while X-ray mapping showed the uniform coverage of the terbium on the graphene oxide particles.

Figure 6: TEM images of (rt) europium graphene oxide, (ctr) graphene oxide and (l) terbium graphene oxide on a holey carbon grid.

Raman analysis was performed on each of the lanthanide nitrates, the lanthanide graphene oxide materials and the graphene oxide itself. The raman was primarily used to verify the binding of the graphene oxide with the lanthanide. The raman spectra also confirmed the graphene oxide maintained its structure within the lanthanide based materials as indicated by the bands at 1400 cm^-1 (D band) and 1600 cm^-1 (G band) region of the spectra. The raman spectra of the terbium species are shown in Figure 7. In addition to the graphene oxide characteristics, the raman also indicated that there is still evidence of some residual nitrate co-ordinated within the materials.

Figure 7: Raman images of terbium nitrate, graphene oxide and terbium graphene oxide.

The ability to co-ordinate luminescent lanthanides with graphene oxide has been established, providing a new set of building blocks for future generations of nanomaterials. The combination of mechanical strength and electronic properties provide by the graphene oxide enhanced by the strong luminescence and magnetic properties of the lanthanides will provide materials for sensors and displays that can be incorporated into fabrics, battery materials and fuel cells with built in residual life indicators, and many other applications. The inclusion of several lanthanides in this study will allow the materials to be much more tunable to color, and magnetism that provide a wider range of chemical applications that could expand the fields of supramolecular chemistry.

The authors would like to thank the Edgewood Chemical Biological Center (ECBC) and the Surface Science Initiative for financial support. Additionally we would like to thank Dr. Erica Valdes (ECBC) for SEM support and Dr. Alex Balboa (ECBC) and Dr. Greg Mogilevsky (Booz Allen Hamilton) for review and useful discussion.