Elements Covers

Tripping the Light Fantastic: Organoboron Compounds

Small molecules containing boron can play all sorts of roles in chemistry, biology and materials science. Molecular boron compounds display a wide range of unusual and fascinating structures, and their chemical reactivity can be very different from that of boron’s next-door neighbour carbon. Some of the reasons for this will be considered and illustrated through applications in energy, medicine and new materials. The boron dipyrrins, also known as BODIPYs, are a prime example. They are strongly fluorescent when excited by illumination and are widely used as fluorescent tags in biology and as biosensors. More recently, they have been studied for their energy transfer properties in light-harvesting applications.

DOI: 10.2138/gselements.13.4.255

Keywords: boron, boranes, fluorescent sensor, sugar sensor, BODIPY

Introduction

Boron occurs in nature largely as borate minerals. Borates in their simplest form occur as calcium or magnesium borates or borosilicates and are dominated by very strong B–O bonds, which at 536 kJ mol−1 are amongst the strongest of the element–element single bonds. These strong bonds also dominate the aqueous chemistry of boron. Looking more broadly at the roles that boron can play in the modern world, we find that small molecules containing boron have a wide range of applications in chemistry, biology and materials science. These applications all have in common that boron is bonded to elements other than oxygen, which opens not only a world of chemical transformations but also very different properties and applications than those of the simple borates that feature in roles as prosaic as washing powder and ant killer. Millions of tons of borates and their derivatives are used worldwide each year, much of it in glass and ceramics, as well as in the oil industry and agriculture (Ritter 2016). Borate may have had a key evolutionary role in stabilizing ribose, critical to forming prebiotic ribonucleic acid (RNA). Boron is an essential trace element in both plants and humans (Furukawa and Kakegawa 2017 this issue).

Polyhedral Boron Clusters

To begin with, consider boron bonded only to itself: in other words, pure, elemental boron. Although it can exist in several amorphous and crystalline phases, the most fascinating to chemists are the a- and b-rhombohedral phases, which contain icosahedral B12 clusters. Although the isolation of pure phases of boron has been dogged by history and controversy, a remarkable set of boron hydride clusters are based on a whole range of polyhedral structures. The boron hydride cluster B12H122 resembles the icosahedral units in rhombohedral boron but with each boron capped by a hydrogen atom, allowing this cluster anion to be isolated as a molecular species (Fig. 1). A range of other boron cluster anions with related formulae BnHn2− (n = 4–12) are known and all are high-symmetry clusters based on deltahedra (regular polyhedra with triangular faces). Neutral boranes with generic formulae BnHn+4 and BnHn+6 have structures based on the same deltahedra but with one or two of the vertices missing. When we compare these boranes to the similar series of binary hydrides—CnH2n+2 (alkanes), CnH2n (alkenes) and CnH2n−2 (alkynes)—we find remarkable differences. Although boron and carbon are immediate neighbours in the periodic table, hydrocarbons are typically linear and branched chains or rings, whereas boron hydrides exist as polyhedral clusters (Fig. 1). Why is this?

  • Facebook
  • Twitter
  • Google+

Figure 1. Whereas hydrocarbons are simple chains or rings with terminal C–H bonds, boranes are clusters with a mixture of terminal B–H and B–H–B bonds.

It all comes down to the electron-deficient nature of boron, which has three valence electrons compared to carbon which has four. Carbon can form four bonds to achieve a closed shell octet, whereas boron can form only three bonds and accumulate six electrons. This is two short of the stable octet configuration. It can make up for its electron shortage by sharing the available electrons as much as possible, leading to curved and, ultimately, closed polyhedral structures. The molecular orbital (MO) theory description of the bonding in these clusters involves MOs both on the surface of and within the cluster, which have contributions from many boron atoms thereby allowing maximum sharing of the available bonding electrons. The MOs in simple hydrocarbons, on the other hand, can generally be described as having contributions from just two atoms, forming simple electron-pair bonds. The boron hydrides just don’t have enough electrons to go around. Even hydrogen gets into the sharing act: hydrocarbons contain simple C–H single bonds whereas boranes contain both simple B–H bonds and unusual B–H–B bridging hydrides, structures that cannot be explained by conventional bonding theories. Harvard chemist William Nunn Lipscomb (1919–2010) won the 1976 Nobel Prize in chemistry for “his studies on the structure of boranes illuminating problems of chemical bonding”. The use of boranes as reagents for chemical transformation was also recognised by a Nobel Prize awarded to Herbert Charles Brown (1912–2004) in 1979.

Applications of Boron in Biology and Medicine

Boron Neutron Capture Therapy

Borane clusters have potential applications, as yet clinically unrealised, in medicine. The stable boron isotope 10B has a very high neutron capture cross section for capture of thermal neutrons, leading to an excited 11B nucleus which decays to a high-energy a particle (4He) and a 7Li nuclei: 10B + n → [11B]* → a + 7Li + 2.31 MeV. If these energetic decay products are generated close to a tumour then ionisations occur which damage the tissue. This therapy, known as boron neutron capture therapy (BNCT) requires a high concentration of boron to be localised in the vicinity of a tumour. Borane clusters conjugated to a tumour-targeting group have been investigated for this. The simplest of these is BSH (B12H11SH2−)— others contain C2B10H11 or C2B9H10 clusters attached to nucleosides, porphyrins or sugars as the targeting agents. Good progress is being made with these, although, as with any new therapeutic agent, issues with solubility, toxicity and efficacy must be addressed along with the stability and accessibility of the neutron source (Luderer et al. 2015; Satapathy et al. 2015).

Boronic Acids

The only drug containing boron that is in current clinical use is bortezomib, approved in 2003 for the treatment of multiple myeloma and non-Hodgkin’s lymphoma. This compound is essentially a tripeptide containing pyrazinoic acid, phenylalanine and leucine with a boronic acid instead of a carboxylic acid at the “C-terminus”. This is the first example of an organoboron compound encountered in this discussion and it contains one B–C bond and two B–OH bonds. One of the best-known uses of boronic acids are as sugar sensors (Hansen and Christensen 2013). A particular example of this is discussed later. A further, very important use of boronic acids is in Suzuki–Miyaura coupling, the most commonly used carbon–carbon bond-forming reaction in the pharmaceutical industry and also widely used for polymerisation reactions (Sakamoto et al. 2009; Blakemore 2016). Akira Suzuki won the 2010 Nobel Prize in chemistry for the development of this important chemical transformation.

Fluorescent Boron and the BODIPYs

The boron dipyrrins, abbreviated to the ‘BODIPYs’, are a group of small, highly fluorescent molecules based on the parent molecule (the core of all BODIPY structures) 4,4-difluoro-4-bora-3a,4a-diaza-s-indacene, which contains a BF2 core bonded to a dipyrrin group (Fig. 2A). They are widely used as dyes, notable for their small Stokes shift, high fluorescence quantum yields, and sharp excitation and emission peaks and high brightness. In Figure 2B, a sample of a typical BODIPY shows that the solid is orange-red but the solution version of the

  • Facebook
  • Twitter
  • Google+

Figure 2. (A) Boron dipyrrin (BODIPY) structure with the dipyrrin core highlighted in green. (B) BODIPY sample in the solid state (crystals on flask wall) and in solution.

compound is a vividly fluorescent yellow-green, even to the naked eye. BODIPY dyes absorb light (a photon of energy) at a characteristic wavelength (forming a transient excited state) and the fluorescence occurs when this energy is re-emitted as light. The emitted light (fluorescence) has a longer wavelength than the absorbed light: the difference is the Stokes shift. BODIPY dyes serve as fluorescent indicators for pH, metal ions, anions, biomolecules, reactive oxygen species and other chemical reactions and physical phenomena (Boens et al. 2012).

Fluorescent Tags for Biomolecules

The parent BODIPY is a scaffold which is modified by appending substituents to the dipyrrin periphery to improve stability and solubility and to tailor the emission properties (Kowada et al. 2015). From this has developed a rich synthetic chemistry for the production of derivatives used in a wide range of applications. BODIPY dyes can penetrate cell membranes and can be used for cell and in vivo imaging. Their fluorescence response can switch between ON and OFF, depending on the mechanism of interaction between the surroundings and the BODIPY excited state. A typical architecture is BODIPY–linker–chelator, in which the chelator is designed to recognise the target analyte. In a reductive photoinduced electron transfer (PET) mechanism, the chelator acts as an electron donor and the excited state BODIPY as an electron acceptor. When the chelator is free, this electron transfer quenches (‘turns off’) the BODIPY fluorescence (Fig. 3A). However, when the chelator recognises the target, it no longer quenches the BODIPY and the fluorescence is turned ON (Fig. 3B). When the donor/acceptor properties are reversed, then oxidative PET can result, and fluorescence is turned OFF when the chelator binds to the target. A second mechanism is the Förster-resonance energy transfer (FRET), which occurs by energy (rather than electron) transfer between the BODIPY and a donor or acceptor.

  • Facebook
  • Twitter
  • Google+

Figure 3. Photoinduced electron transfer (PET) mechanism for a turn-ON boron dipyrrin (BODIPY) sensor. (A) The free chelator quenches BODIPY fluorescence via photoinduced electron transfer. (B) When the target is bound to the chelator, then PET is inhibited and fluorescence turned ON. Commercially available BODIPY probes have been developed for fluorescence microscopy of lysosomes (subcellular organelles). The green- and red-labelled NMe2 chelator groups target the protons (H+) in acidic compartments in cells. (C) LysoTracker® Green, a green fluorescent dye that stains acidic compartments in live cells. (D) LysoTracker® Red, is very similar to LysoTracker® Green, but the extra small ring attached to the BODIPY (labelled red) changes the fluorescence wavelength from green to red.

BODIPYs can be used for single molecule and super-resolution microscopy, a technique for visualizing individual proteins and biomolecules. Examples are the commercially available LysoTracker® Green (excitation/emission maxima 504/511 nm) and LysoTracker® Red (excitation/emission maxima 577⁄590 nm) probes, both developed for fluorescence microscopy of lysosomes (subcellular organelles). They can be modified to fluoresce at different wavelengths by variation of the substituents on the BODIPY core, and are sensitive to pH. The dyes are weakly fluorescent in neutral or alkaline conditions, but their fluorescence is greatly enhanced in the acidic environment of lysosomes when the terminal nitrogens on the side chain (highlighted in green and red in Figs. 3C and 3D) are protonated (equivalent to the chelator in Fig. 3B binding to its target, in this case H+). The second red highlight in LysoTracker® Red illustrates how addition of the electron-rich pyrrole unit to the dipyrrin framework dramatically shifts the absorption and emission wavelengths and, hence, the colours (Fig. 3).

Sugar Sensing

The B–F bonds in BODIPY are very strong and, as a result, chemically inert. They stay intact in all the applications for which BODIPY dyes are used. Modifications to BODIPYs by attaching groups to modify the fluorescence properties or by targeting the analyte occur on the dipyrrin carbon backbone. Further modifications could be induced by replacing the fluorine (F) substituents on boron, although the extremely strong B–F bonds make this chemically challenging. However, by using an appropriately strong Lewis acid (see below for a discussion of these) to activate the B–F bonds, substitutions can be made, for example with chlorine or oxygen to make Cl-BODIPY or O-BODIPY, respectively. The B–Cl and B–O bonds are more reactive, and with the right substrate will undergo chemical reactions at the boron centre. In this way, the boron itself can be used as the recognition unit for the target analyte.

Boronic acids were mentioned above—they are organoboron compounds containing two B–OH bonds and one B–C bond (denoted “R”). Sugar molecules contain several C–OH groups and, when in water, boronic acid and sugars can form links though B–O–C bonds. If the organic group R on the boronic acid has a “reporter” function, for example fluorescence turning ON or OFF upon the formation of the boronic acid–sugar bonds, then this allows the system to sense the presence or absence of a sugar (Fig. 4).

  • Facebook
  • Twitter
  • Google+

Figure 4. Schematic of a boronic acid sugar sensor. When boronic acid is linked to sugar OH groups, the organic group on the acid undergoes a measurable change (fluorescence) that can be used to detect the presence or absence of sugar.

In very recent research from our own laboratory, we have combined the concept of boronic acids as sugar sensors with substitution at the boron centre in O-BODIPY to make direct BODIPY–sugar conjugates (Fig. 5). Using glucose as the sugar, we can isolate several examples containing glucose in different ring forms and with 1:1 and 2:1 BODIPY:glucose ratios. These retain their excellent fluorescence properties. We are currently investigating their possible applications for recognising and targeting specific sugars and to be developed as probes for polysaccharide structure and function (Liu et al. 2016).

  • Facebook
  • Twitter
  • Google+

Figure 5. Schematic illustrating the boron dipyrrin (BODIPY)–sugar concept. When F-BODIPY links to sugar OH groups, such as glucose, the resulting O-BODIPY complex fluoresces.

Boron in Energy and Materials

Light Harvesting

BODIPYs were discussed above for their applications as fluorescent sensors. The mechanisms by which BODIPY fluorescence is modulated are typically by photoinduced electron transfer (PET) or by Förster resonance energy transfer (FRET – energy transfer between two light-sensitive molecules). In systems which are tailored for FRET, BODIPYs have been applied to light harvesting. Their good light-absorption properties, coupled with their ability to transfer energy via FRET, make them useful for incorporation into molecular systems engineered to capture light energy and transform it to an alternative form in which it could be used or stored. The important steps in these light-harvesting antenna systems are the absorption of light by an antenna, followed by energy transfer, which in turn drives electron transfer from a donor to an acceptor. This produces an electron–hole pair, which, if it has a sufficiently long lifetime, can be drawn off into an external electrical current. The photosynthetic reaction centre in chlorophyll is the source of inspiration from which this approach is derived. In an elegant example of this principle, a molecular construct features BODIPY units as the antennae, a zinc porphyrin (closely related to the iron porphyrin or heme unit found in hemoglobin) as the electron donor, and a C60 fullerene as the electron acceptor (Fig. 6) (Maligaspe et al. 2010). The BODIPYs absorb light and they transfer this energy to the porphyrin, which drives electron transfer to the fullerene. This is just one example of the ways in which BODIPY ‘sensitizers’ are being used as antennae in light-harvesting systems—and there are many more.

  • Facebook
  • Twitter
  • Google+

Figure 6. In this BODIPY–porphyin–fullerene molecular system for light harvesting, the BODIPY molecule acts as a sensitizer, absorbing a photon of light. This light energy is transferred to a zinc porphyrin electron donor and is used to drive electron transfer to the fullerene electron acceptor. This molecular chain is an example of how light energy can be converted to a flow of electrons and, ultimately, an electrical current.

Boranes as Fuels

We have considered above the boron hydrides, or boranes. Boranes share with the hydrocarbons very high heats of combustion, forming boron oxide, B2O3 + H2O, similar to the formation of CO2 + H2O upon combustion of hydrocarbons. For example, burning ethane (C2H6) produces 1,560 kJ/mol of energy, while burning diborane (B2H6) produces 2,138 kJ/mol. Boron hydrides can achieve higher energy densities than hydrocarbons, and so-called “zip fuels”—a family of jet fuels containing boranes as additives—were investigated during the cold war era. However, both the fuels and the exhaust were toxic and the boron carbide by-products of incomplete combustion caused problems with turbine blades. Their use was discontinued. The lighter members of the borane family, for example diborane (B2H6), are pyrophoric, which means that they spontaneously combust on contact with air. This property has been exploited in the case of triethylborane (an organoboron derivative), which was used to ignite the slippery and high flash-point jet fuel JP-7 used in the SR-71 Blackbird’s jet engine.

Hydrogen Storage

One of the most exciting possibilities of a post–fossil fuel world is the use of hydrogen as a fuel. Currently, hydrogen is transported and stored as a compressed gas. The steel cylinder used to store compressed hydrogen gas adds a lot of weight to the total system (for example, on-board storage in a hydrogen-powered car), poses the risk of working with high pressures, and cannot achieve an energy density as high as liquid gasoline. A safer, more efficient, way of storing hydrogen might be via chemical hydrogen storage. This concept requires a lightweight, hydrogen-rich material that can release hydrogen on demand under mild conditions, and be recharged by reversing the process, i.e. absorbing hydrogen to reform the hydrogen-rich phase. Ammonia borane (BH3–NH3) contains 19.6% hydrogen by weight, releases hydrogen under mild conditions (temperatures around 100 °C) and has been intensively investigated as a chemical hydrogen storage material (Heldebrant et al. 2008).

Boron Nitride

If all the hydrogen is removed from ammonia borane, then the resulting chemical product is boron nitride (BN). This material is fascinating for several reasons, but first let’s think about it from fundamental principles. Boron, carbon and nitrogen have 3, 4 and 5 valence electrons, respectively. Hypothetical C2 and BN fragments each have 8 valence electrons. Extending this analogy, bulk carbon and bulk boron nitride should be isoelectronic and show similar chemical and materials behaviour. This turns out to be exactly the case. Just as carbon can exist as diamond and graphite allotropes, so boron nitride can exist in two similar phases (Fig. 7). The cubic form (sphalerite structure) is isostructural with diamond, and although not quite as hard, has better thermal and chemical stability. Hexagonal BN has a graphite structure and is a good lubricant—again, its higher stability renders it useful under conditions too harsh for graphite itself. It has a high sheen and is used as an additive in cosmetics to give them a pearly lustre. An exciting recent development in carbon chemistry is the production of carbon nanotubes—these are sheets of graphitic carbon rolled to make nano-sized tubes, and they have many applications. Just as might be expected from the relationship between BN and carbon, BN nanotubes are also known, but have rather different properties. Carbon nanotubes are metallic or semiconducting, while BN nanotubes are insulators. Buckminsterfullerene (C60) is a cluster of 60 carbon atoms with the same symmetry as a soccer ball; BN fullerenes are also known, but are often observed as multiwalled, concentric structures (Zhi et al. 2008). All these new nanostructures are the subject of intense research as chemists and materials scientists work to understand their properties and develop new applications.

  • Facebook
  • Twitter
  • Google+

Figure 7. Cubic and hexagonal phases of boron nitride (BN) and a multiwalled BN nanotube. Reproduced from Zhi et al. (2008) with permission of the Royal Society of Chemistry.

Boron Compounds in Catalysis: Frustrated Lewis Pairs

The electron deficiency typical of boron compounds manifests itself as strong Lewis acidity. Lewis acidity is a broader concept than simple proton acidity (also known as Brønsted acidity). Brønsted acid–base concepts define an acid (like hydrochloric acid, HCl) as a proton donor, and a base (like hydroxide ion, OH−) as a proton acceptor. The Lewis acidity concept recognises that a proton donor is also an electron pair acceptor and uses this as a definition of acidity. Thus, a Lewis acid is an electron pair acceptor and a Lewis base an electron pair donor. Trivalent boron compounds (denoted BX3 where X is any atom or group attached to boron) are very good Lewis acids. They contain only a sextet of electrons (three B–X bonds each requiring two electrons) and have a high driving force to accept a further pair of electrons (a Lewis base electron pair donor) to achieve an octet. A simple example is the reaction between BF3 (Lewis acid) and NH3 (Lewis base) to form the BF3–NH3 complex which contains a new B–N bond. Ammonia borane (BH3–NH3), described above in the discussion on hydrogen storage, is an example of a Lewis acid–base complex.

Recent work in this sphere has extended this concept to what are termed “frustrated Lewis pairs” (FLPs). A boron Lewis acid with very bulky groups attached (BR3 where the R groups are large and take up a lot of space) coupled with a similarly bulky Lewis base results in an FLP. The bulky molecules want to form a complex with a new bond to boron but are prevented from doing so because the groups around each partner take up too much space. The FLPs turn out to be very reactive combinations – the Lewis acid and Lewis base centres want to react with each other, but they can’t. So, instead the FLPs will react with other small molecules.

  • Facebook
  • Twitter
  • Google+

Figure 8. Illustration of “frustrated Lewis pairs”. The bulky groups on boron and phosphorus mean the boron of the Lewis acid (C6F5)3B cannot get close enough to the phosphorus of the Lewis base PtBu3 to form a B–P bond, resulting in a “frustrated Lewis pair”. If H2 is present, the pair reacts with hydrogen, with boron accepting a hydride (H) and phosphorous accepting a proton (H+). Shown is the resulting ion pair (left) [(C6F5)3BH] and (right) [HPtBu3]+ formed from the reaction of (C6F5)3B and PtBu3 with H2. From Welch and Stephan (2007).

In the example shown in Figure 8, the boron of the Lewis acid (C6F5)3B cannot get close enough to the phosphorus of the phosphine Lewis base PtBu3 (a phosphorus atom bonded to three branched organic groups) to form a complex with a B–P bond, but instead the FLP reacts with hydrogen, with the boron accepting a hydride (H−) and phosphorus accepting a proton (H+) (Welch and Stephan 2007). This uncoupling of H2 into H− + H+ is difficult to accomplish by other means. Although it burns readily (explosively!), hydrogen gas (H2) is a relatively inert molecule and must be activated towards chemical reactivity, usually by catalysis. A catalyst is not consumed in the reaction but provides a pathway for the reaction to occur. Hydrogenation, or transfer of H2 to other molecules to produce new chemical compounds, is of enormous importance in all kinds of chemistry ranging from drug synthesis to food processing. Typical catalysts are based on heavy elements such as palladium, which are useful on a laboratory scale but expensive and unsustainable on an industrial scale. Remarkably, the FLP/H2 system using boron compounds as the Lewis acids turn out to be excellent hydrogenation catalysts (Stephan and Erker 2015). The significance of this is that expensive palladium catalysts could be replaced by a much cheaper and more readily available element like boron, avoiding the use of heavy metals altogether.

CONCLUSIONS

Modern boron chemistry, particularly molecular organoboron compounds, is coming of age with applications in new materials, sensing, medicine, catalysis, and energy all developing apace. In a world where there is increasing focus on future sustainability, one important driver is the development of chemical systems that do not rely on heavy metals for important transformations and processes. Boron, one of the lightest elements in the Periodic Table, fits the bill for this very well. The examples above illustrate some of the possibilities being pursued in current boron research. Some ideas have been around for a while (boron neutron capture therapy) whereas others are very new (frustrated Lewis pairs for metal-free catalysis and BODIPY compounds in light harvesting).

Tripping the light fantastic – the important applications of boron materials described above derive either from boron’s properties as a light-weight element or its ability to form fluorescent (light-emitting) organoboron compounds. Without a doubt, there will be more to come.

REFERENCES

Blakemore D (2016) Suzuki–Miyaura Coupling. In: Blakemore DC, Doyle PM, Fobian YM (eds) Synthetic Methods in Drug Discovery, Volume 1. The Royal Society of Chemistry. pp 1-69

Boens N, Leen V, Dehaen W (2012) Fluorescent indicators based on BODIPY. Chemical Society Reviews 41: 1130-1172

Furukawa Y, Kakegawa T (2017) Borate and the origin of RNA: A model for the precursors to life. Elements 13: 261-265

Hansen JS, Christensen JB (2013) Recent advances in fluorescent arylboronic acids for glucose sensing. Biosensors 3: 400-418

Heldebrant DJ, Abhi Karkamkar A, Linehan JC, Autrey T (2008) Synthesis of ammonia borane for hydrogen storage applications. Energy and Environmental Science 1: 156-160

Kowada T, Maeda H, Kikuchi K (2015) BODIPY-based probes for the fluorescence imaging of biomolecules in living cells. Chemical Society Reviews 44: 4953-4972

Liu B and 7 coauthors (2016) Lighting up sugars: fluorescent BODIPY-gluco-furanose and -septanose conjugates linked by direct B–O–C bonds. Organic and Biomolecular Chemistry 14: 5205-5209

Luderer MJ, de la Puente P, Azab AK (2015) Advancements in tumor targeting strategies for boron neutron capture therapy. Pharmaceutical Research 32: 2824-2836

Maligaspe E and 6 coauthors (2010) Electronic energy harvesting multi BODIPY-zinc porphyrin dyads accommodating fullerene as photosynthetic composite of antenna-reaction center. Physical Chemistry Chemical Physics 12: 7434-7444

Ritter SK (2016) Boron chemistry branches out. Chemical and Engineering News 94: 20-23

Sakamoto J, Rehahn M, Wegner G, Schlüter AD (2009) Suzuki polycondensation: polyarylenes à la carte. Macromolecular Rapid Communications 30: 653-687

Satapathy R and 5 coauthors (2015) Glycoconjugates of polyhedral boron clusters. Journal of Organometallic Chemistry 798: 13-23

Stephan DW, Erker G (2015) Frustrated Lewis pair chemistry: development and perspectives. Angewandte Chemie, International Edition 54: 6400-6441

Welch GC, Stephan DW (2007) Facile heterolytic cleavage of dihydrogen by phosphines and boranes. Journal of the American Chemical Society 129: 1880-1881

Zhi CY, Bando Y, Tang CC, Huang Q, Golberg D (2008) Boron nitride nanotubes: functionalization and composites. Journal of Materials Chemistry 18: 3900-3908

Share This