The Quest for the Simplest Possible Organogelators and Some Properties of their Organogels

The search for the simplest structural class of organic molecules capable of gelling large amounts of organic liquids is reviewed with an emphasis on the approach taken by the authors. The properties of several of the gelators, as well as their gels with various organic liquids, are presented. Some challenges to future developments in the field are mentioned.


Introduction
This review 1 focuses on recent research at Georgetown University on low molecular-mass organic gelators (LMOGs, molecules whose molecular masses are ≤ 3000 g mol L -1 ) and their organogels.It describes our efforts to reduce the structural complexity of the LMOGs --leading eventually to the simplest class of organogelators possible --and approaches taken to discern the structures of their gels at various length scales.
"What is an organogel?" is a question that has been addressed without a very satisfactory answer for more than a century.There are several types of organogels, and each requires a definition with somewhat different qualifications.
Unfortunately, the statement by Jordan Lloyd more than 70 years ago, "…the colloid condition, the gel, is easier to recognize than to define" 2 , was prophetic.In fact, many gels are not colloidal!Flory's 3 definition of gels 4 attests to their complexity; it is rigorous, but very difficult to apply on a routine basis.Even gels with polymeric gelators are difficult to define 5 .For the purposes of this review, organogels must be composed of a low concentration (usually ≤ 2 wt%) of an LMOG in an organic liquid and meet two loosely defined criteria: (1) They may be distorted in shape by an applied stress (below a certain limit) but must return to their original form when relieved of the stress.
(2) Although being composed predominately of liquid and being fluid at the microscopic scale, they must appear solid-like macroscopically.
In addition, all LMOG organogels are thermally reversible (that is, they can be cycled repeatedly with their pregelation (sol) phases by heating and cooling) unless there is a chemical reaction that occurs along with the physical changes.
Understanding how LMOG molecules nucleate and assemble in the sol phase and whether the lyotropic structures in the sol resemble those in the gel are keys to learning the mechanisms for organogel formation.Such considerations need not be addressed to understand polymer gels and many hydrogels.In this regard, it is important to distinguish the critical aggregation concentration (CAC; i.e., when LMOGs aggregate in the sol) and the critical gelator concentrations (CGC; i.e., the lowest concentration of LMOG molecules capable of gelling a liquid at room temperature) 6,7 .The aggregation process leading to gelation has been studied by optical spectroscopic methods 7 and, in a few cases, by small angle neutron scattering (SANS) 8,9 , electron spin resonance (ESR) 8 , and atomic force microscopy (AFM) 10 .
LMOGs self-assemble usually via one-dimensional growth modes to form fibers, strands, or tapes which are frequently crystalline.Recent examples of gelator structures based on microplatelets indicate that two-dimensional growth patterns by the nucleating species are also possible 11 .Although it is generally assumed that strong intermolecular forces such as H-bonding, electrostatic attractions, or π-π stacking interactions are necessary to stabilize LMOG assemblies, recent observations 11 have demonstrated that London dispersion forces, alone, can be sufficient.The primary nanoscale objects, regardless of their shapes, join in three-dimensional networks that encapsulate the liquid component and inhibit its flow."Junction zones" 1a between fibers, strands, tapes, or microplatelets provide rigidity to the microstructure.
Many of the solid structures are colloidal in nature and gelation occurs when the individual colloids interact physically while pervading the liquid volume.Regardless, the highly porous superstructure of the linked nanoscale objects immobilizes a large volume of liquid via surface tension and related forces 12 .In addition, a few LMOG gels are reported to be thixotropic 13,14 .

A personalized history of the recent development of LMOG research
ALS and related LMOGs.Our entry into the field of organogels was serendipitous.It resulted from an observation by Y. -C.Lin during a photochemical investigation 15 that small concentrations (typically < 2 wt%) of 3(β)cholesteryl 4-(2-anthryloxy)butanoate (CAB) gelled a wide variety of organic liquids 16 .Initially, we looked upon the gels as an undesirable nuisance!At room temperature in a closed vessel, some CAB gels are stable for years and others separate macroscopically to a solid and a liquid after no more than a few minutes.
"building blocks" within the sol phase combine only along selected faces to create long stacks.In addition, strands of CAB from 1-octanol gels are twisted with a pitch of ca.120 nm while those from n-alkane gels are not twisted.A particularly elegant example of the control of helicity and chirality is found in LMOG tapes based on 12-hydroxyoctadecanoic acid 7d, 18 .
Subsequent investigations have shown that CAB is only one member of a class of molecules with an aromatic, linking, and steroidal part (i.e., ALS molecules), and many of them function as LMOGs 7a,b .Structural changes have been made to each part of the ALS structure, including the stereochemistry at C3 and the nature of the chain at C17 of the steroidal part.2-Substituted-9,10-anthraquinones, cinnamate, N-substituted anilines, 2-naphthyl, 1-pyrenyl, and p-substituted phenyl have been introduced as the aromatic part.The length and functionality of the linker have been modified, also.Of the more than 40 ALS molecules synthesized by us, 19 are able to gel at least some organic liquids.In addition, others have synthesized ALS gelators containing substitiuted azobenzenes 7c , squarines 44h , and stilbenes 44h .
It has been possible to derive several important conclusions concerning the nature of LMOG organogels from these investigations: (1) H-bonding, even when possible, may be absent in LMOG assemblies when other packing contributors (e.g., π-π interactions and London dispersion forces) dominate 7a , (2) charge-transfer interactions within gelator strands can stabilize gels 14b,19 , (3) thixotropy 13 can be induced by adding a small concentration of a second (nongelling) ALS molecule whose size and shape are similar to those of a good ALS gelator 14b , (4) the fraction of ALS gelator within the solid network is dependent on temperature and the solubility of the gelator in the liquid component 7b,20 , (5) the bulk properties of a liquid mixture, rather than the properties of the individual components, determine the dimensions and shape of the gelator assemblies 21 , and (6) subtle changes in molecular shape can alter profoundly the ability of an ALS to gel organic liquids (e.g., 2-CA is an excellent gelator of many types of organic liquids but 9-CA did not gel any of the same liquids 7b ).
Unfortunately, the complexity of the ALS structures has not allowed a clear, broadly applicable link between molecular structure and gelation ability to be established.However, it is clear that LMOG structure alone does not govern gelation ability.Widely applicable correlations between structure and function must be derived from LMOGs that are simpler than the ALS.
With this goal in mind, others 22 and we 7b,14b have examined the gelling abilities of AL molecules (i.e., containing an aromatic group and one "linking" chain, but lacking a steroidal group) with very little success 23 .
The CAB gel superstructure has been characterized in detail 7b .The size of its colloids depends upon the nature of the liquid and the rate at which the sol is cooled to below T gel , the sol↔gel transition temperature.For instance, colloids from gels with 1-octanol are < 10 µm in diameter while those with n-alkanes are more than one order of magnitude larger.As a result, gels are not formed by CAB and n-alkanes when one dimension of the vessel in which they are contained is ≤ 100 µm 17 .The substructures of the colloids are elaborately branched strands whose rectangular cross-sections range from 10 to 20 nm in size and are nearly monodisperse within one gel.The uniform cross-sections in these and other gels 1a suggest that growth beyond certain limits along two axes is prohibited, but essentially unlimited growth is allowed along the third!An attractive hypothesis to explain this observation is that However, several AL 2 molecules (i.e., containing one aromatic and two linker groups) are efficient LMOGs.An example, 2,3-di(dodecyloxy)anthracene (DDOA), gels several organic liquids especially at low temperatures 22 .When the length of its alkyl chains were shortened, the oxygen atoms were removed, the chain lengths were mismatched, or the aromatic part was truncated to naphthyl, the gelating ability of DDOA was either lost or severely reduced.
However, hydrogenation of one of the anthryl rings of DDOA 24 or replacement of it by a phenazine (DNON and DUON) 25a or an anthraquinone (DDOQ) 25b produced LMOGs whose efficiencies are similar to that of the parent.In addition, a fullerene containing AL 2 molecule gels methanol, but only when sols stand undisturbed for protracted periods 26 .Its gelating ability may derive more from its two trimethylammonium bromide groups that terminate the two L chains than from the fullerene A part 27 .

LMOG salts
For this reason, we attempted to discover LMOGs consisting essentially of only an L part (i.e., alkyl chains with minimal functionalization).The simplest LMOG structures known before 1997 11,33 were long, partially fluorinated n-alkanes 34 , but they gel a limited number of liquids and require rather large concentrations to do so.The first approach to further simplifications was suggested by the efficiency of cholesteryl tri-n-alkyl ammonium LS gelators (e.g., CDOAI; Scheme 1) 35 .When the cholesteric group was replaced by another n-alkyl chain, several of the tetra-n-alkylammonium salts were very efficient LMOGs, especially when three or four groups are "long"; in the first examples, octadecyl chains were employed (e.g., 18NBr; Scheme 1) 27 .However, these LMOGs are unstable at elevated temperatures due, probably, to Hofman-type elimination reactions.Replacement of the nitrogen atom of the cationic head group with phosphorus, another Group VA atom, provided much more stable phosphonium salts 36 whose gelation efficiencies are somewhat different from the corresponding ammonium salts 37 .
The gelling properties of salts with four equivalent chains, H(CH 2 ) n ) 4 Y + X -(nYA, where n, the number of carbon atoms in each alkyl chain, is varied from 7 to 18, Y is N or P, and A is Cl, Br, I or ClO 4 ), will be discussed in some detail.The influence of n and Y are emphasized; we have not been able to discern a correlation between the size or type of X group and the ability of an nYA salt to be an LMOG.
Comparisons between identically prepared nYA gels demonstrate that the ammonium salt LMOGs have higher T gel values, are stable for longer periods in sealed vials at room temperature, and require lower CGCs than the corresponding phosphonium salts.Stronger N + A -than P + A - interactions are believed to be largely responsible for these observations.Polarizability differences between ammonium and phosphonium cationic centers do not appear to be important since the valence shell electrons about the hetero-atom of the nY + parts are well shielded from the anion by the four alkyl chains.However, greater ionic interactions can result from tighter packing of the α and β Some LS molecules (i.e., containing a "linking" chain and a steroidal group) have also been examined.Although several cholesteryl alkanoates do not gel simple organic liquids 7b , cholesteryl laurate gels some silicone oils 28 , and several steroidal amines and their salts 27 have been shown to be efficient LMOGs.Even some S molecules (i.e., consisting of only a steroidal group) are good LMOGs.Examples include dihydrolanosterol 29 and lithocholic acid salts 30 .Gels of some 17-azahomosteroids of isoandrosterone 31 have been investigated in great detail by Terech and coworkers using scattering and rheological techniques 32 .In spite of the simplifications introduced by removal of the A, S, or both parts of the ALS gelators, it is still difficult to correlate their specific structural changes with variations in the properties of their organogels.methylene units (i.e., those nearest nitrogen).Since N-C covalent bond distances (~1.53 Å) are shorter than P-C ones (~1.81 Å), an N + center can approach its anions more closely than can P + 38 .Additionally, the larger inductive effects 39 of nitrogen make the α and β methylene hydrogen atoms of the nN + cations more acidic (i.e., they bear a larger partial positive charge) than those of the nP + cations; stronger hydrogen-bonding and, therefore, shorter hydrogen-anion contact distances are expected for the ammonium salts.
From single crystal x-ray diffraction studies on several of these salts, the separation between cationic and anionic centers minus the van der Waals radius of nitrogen or phosphorus, as appropriate, is always smaller for the nNA salts than the corresponding nPA salts 38 .However, we are reluctant to draw a strong conclusion from this observation because the morph of an LMOG in its gels may differ from that in solids obtained by bulk recrystallizations 40 .
The magnitudes of the London dispersion forces (that supplement the stronger Y + A -electrostatic interactions) are responsible for greater gelation efficiencies by nYA salts with longer chains.The nNBr salts exemplify the trends.Due to its high solubility in all of the liquids tested, 7NBr was unable to form a gel.Relatively high concentrations (~0.2 mol L -1 ) of the slightly longer 10NBr gelled hexadecane (stable at room temperature for ≤ 2 days) and glycidyl methacrylate, but not benzene, 1-octanol, carbon tetrachloride, or styrene.Generally, as n of the nNBr increased, the CGC decreased and T gel values increased.In addition, the melting temperatures of the salt gelators parallel T gel of their gels.Gels from salts with shorter chains usually had wider T gel transitions because they are solubilized more gradually as temperature is increased than salts with longer chains.Consequently, a clear assignment of T gel was not always possible for salt gels with shorter chains.
In addition to increasing T gel and lowering the CGC, longer chain lengths of salt gelators increase the periods of gel stability.For instance, some hexadecane gels with 16NBr and 18NBr gelators have persisted for more than 2 years in closed vials at room temperature!By contrast, hexadecane/12NBr gels with the same wt% composition survived less than 2 days.Gels of the other liquids in Table 1 follow qualitatively the same trend.CCl 4 /16NBr and CCl 4 /18NBr gels, which require > 5 wt% gelator, persisted less than 1 week.They became yellow with time, indicating that some decomposition had occurred.
Thermograms of nNBr gels (Figure 1) illustrate further the dependence of alkyl chain length on gelation ability.Salts with the longest chains provide gels with the lowest CGCs and the highest T gel values, regardless of cooling protocol.If the packing of gelator molecules in gel strands is microscopically separated into lipophilic and lipophobic regions, as in their bulk solid states 38 , longer alkyl chain lengths will increase the aggregate stabilities as a result of larger London dispersion forces.1) nNBr/hexadecane gels at 0.01 and 0.08 mol L -1 gelator concentrations.The values of n for each concentration are in the order 12→14→16→18 from bottom to top 37 .
The dependence of cooling protocol on T gel and gelation is evident in Figure 2. T gel is higher for slowly cooled gels.However, gelation is facilitated by quickly cooling sols.Dilute, fast-cooled gels of nYA LMOGs are more translucent than slow-cooled ones, also.The networks of the former are more intricate and have larger surface/ gelator-mass ratios that allow them to entrap a larger volume of liquid by surface tension 37 .For example, fastcooled sols of 18PBr or 18PI in aromatic liquids provide translucent gels with a bluish tint (Tyndall effect), but the same sols yield white and opaque gels when slowly cooled.
quaternary analogue, but it is able to gel some organic liquids 41 .Methyl-di-n-octadecylamine (MeN), in which one of the octadecyl chains has been truncated to methyl, is less efficient than 3N, but truncating the methyl group further to hydrogen as in di-n-octadecylamine (2N) increases gelator efficiency.2N can act as both a donor and acceptor of H-bonds; the presence of H-bonds was confirmed in strands of 2N gels by infrared spectroscopy 41 .3N, MeN, and di-n-tetradecylsulfide (2S), another single hetero-atom LMOG, may be less efficient than 2N because they are able to accept H-bonds only.However, 1octadecylamine (1N), capable of being an H-bond donor and acceptor like 2N, did not gel any of the liquids tested.We suspect that the molecular packing arrangement of its solid may not be amenable to formation of gels and that the presence of some octadecylammonium octadecylcarbamate, invariably present when 1N is exposed to air 42 , may catalyze the nucleation of morphs that are not amenable to strands or other gel-related motifs 41 .
Unfortunately, the organization of these LMOGs in their gel strands is not known.However, the positions and appearances of the IR absorption bands for the N-H stretch of 2N in gels suggest that the same solid morph may be responsible for its bulk solid at least in siloxane gels.Since the melting temperature of neat 2N is the highest of the five single hetero-atom gelators examined, its intermolecular interactions may be strongest, at least in the bulk solid phase.On a per octadecyl chain basis, the enthalpy (and entropy) of melting of the four amines as bulk solids follow the order: 1N (73.3 kJ mol-chain -1 (226.5 J mol-chain -1 K -1 )) > 2N (60.3 kJ mol-chain -1 (182.9J molchain -1 K -1 )) > 3N (45.3 kJ mol-chain -1 (141.5 J mol-chain -1 K -1 )) > MeN (39.0 kJ mol-chain -1 (125.5 J mol-chain -1 K -1 )).The higher enthalpy and entropy per octadecyl chain for 1N and 2N are consistent with strong hydrogenbonding interactions that are not possible in the other (cont.)Table 1.

Concentration
Cooling  b Cooling protocols for samples in sealed tubes: slow -submerged in initially boiling water until room temperature achieved; moderate --removed from the boiling water and placed in the air; fast -removed from the boiling water and placed under a stream of water at 20 o C. c MMA = methyl methacrylate.d GMA = glycidyl methacrylate.g = gel formed but T gel not determined; pg = partial gel.Where gels did not form: s = solution; p = solid + liquid.The influence of cooling rates of nNBr/styrene sols on the T gel values of their gels: n = 18 (s), 16 (q), 14 (v), 12 (x, ); fast ( ), moderate (-------) and slow (•••••••) cooling; see Table 1.Averages are reported when ranges of melting were noted.The value is the same for fast, moderate and slow cooling 37 .

LMOGs with one hetero-atom (Scheme 2)
Each of the ammonium or phosphonium LMOGs has a minimum of two hetero-atoms-the cationic center and its anion.Removal of one of the salt chains leads to molecules with only one hetero-atom and three chains.Due to the instability of phosphines in air, the gelling characteristics of only the amines were explored.As expected, tri-noctadecylamine (3N), a tertiary amine with three long n-alkyl chains, is a much less efficient LMOG than its Weight % nNBr gelators.This is most apparent when comparing the structurally comparable gelators, 2N and MeN, where the transition enthalpy and entropy of the H-bonding molecule are 55 % and 46 % greater, respectively, than those of the methylated one.At equal molar concentrations of single hetero-atom LMOGs, T gel of 1-alkanol gels decreased as the chain length of the liquid was increased.Not inconsequentially, these gelators are more soluble in the longer alcohols, and 2S, MeN, and 3N remained solubilized when placed in nalkane liquids (that mimic the polarity effect of making alkanol chains infinitely long).

n-Alkanes, LMOGs with no hetero-atoms
Stabilization of the gel assemblies of 2S, a molecule whose chain length equals that of 1N, but which cannot donate H-bonds, must rely heavily on van der Waals forces 41 .On that basis, we wondered whether n-alkanes with long chains might serve as LMOGs.To our amazement (and joy), several organic liquids, including shorter n-alkanes, have been gelled thermoreversibly by low concentrations of longer n-alkanes (Cn with n = 24-36)! 11elator efficiency increases with alkane chain length.
Some comparisons can be made between n-alkane and single hetero-atom n-alkane LMOGs.2S is C28 with an S atom inserted at its center, and the structure of 2N is C36 with an N (and hydrogen) atom at its center.Both 2S and C28 are much more soluble than their longer chain analogs.We have been able to gel only alcoholic liquids with 2S and C28 above room temperature, and 2S is unable to gel several alcohols that can be by C28.However, neither molecule is as efficient an LMOG as its longer chain analogs.Both C36 and 2N form gels that are stable above room temperature with most of the liquids tested.The more polar 2N has lower T gel and periods of stability in alcoholic liquids and the less polar C36 has lower T gel and periods of stability in non-polar liquids.Again, these trends are related to the solubility of the LMOGs in the liquid components.Additionally, 2N is more soluble and formed less stable gels in liquids capable of donating or accepting hydrogen bonds than C36.

The structures of C36 organo gel assemblies at Ångstrom to micrometer distance scales
Recently, we determined the first complete description of the packing of an LMOG in its organogels 43 .Prior attempts have not been completely successful for several reasons, including the polymorphism of many organogelators and the difficulties in making diffraction quality single crystals of others.Structural information on LMOG packing at the molecular level has been inferred from conventional techniques, such as nuclear magnetic resonance spectra 44a , semi-empirical calculations 7c or assumptions that the gelator superstructure is the same as the xerogel morph 44b or the morph analyzed by single-crystal X-ray diffraction (in the relatively few cases where this is available) 44c,d,e .At the supermolecular scale, information can be gained by SANS, SAXS, electron microscopy, and atomic force microscopy 7b,44f,g,h .Unfortunately, extrapolation from the supermolecular to the molecular scale using any of these methods is not definitive.
Previously, a method was devised in our group to provide molecular packing information of LMOGs within their gel superstructures by relating the X-ray diffraction (XRD) pattern of a gel to that of its neat solid phase 40 .When the two patterns are the same and the single-crystal structure of the LMOG is also available, packing in the gel is determined unambiguously.In its first application 40 , the method demonstrated that the morph of the LMOG, 3(β)-cholesteryl anthraquinone-2-carboxylate (CAQ), exists in gel strands in the same morph as the solid derived from the melt, but different from the morph derived from bulk crystallization.Unfortunately, molecular structural information is available only for the latter.

Conclusions and prospects for the future
Careful scrutiny of the literature indicates that engineers involved in fuel transport have been aware of a phenomenon like gelation for many years, and view it as a nuisance to crude oil flow 47 .In cold climates, diesel-burning automobiles are equipped with heaters for the fuel tanks in order to avoid gelation by the long n-alkane components.Terms such as 'wax-appearance' and 'cloud point' are used in the fuel community to describe what others 33 and we have concluded are gelation processes.
n-Alkanes are structurally the simplest LMOGs possible and their gels with n-alkanes as liquids are the simplest organogels that can be made.The existence of these gels demonstrates that London dispersion forces alone can provide solid networks whose strength is sufficient to immobilize liquids against the pull of gravity.The currently accepted paradigm 1a (formulated in part by the elder and less wise of the two authors) does not predict the existence of such organogels.
Over more than one decade, we have progressed (or regressed?)from very complex ALS structures to the simplest LMOGs possible, n-alkanes.Thus, we are finally in a position to examine systematically the relationship between structural changes in an LMOG and the stability of its gels.However, many of the remaining questions concerning why LMOG gels form can be answered only through investigations of the processes leading from sols to the gels, including the nucleation of LMOG aggregates and the assembly of the aggregates into the colloidal superstructures.Despite the progress made, many challenges lie ahead.Using this same approach, we have identified the single morph (of the four that are known 45 ) of the LMOG, C36, that is responsible for gelation of several liquids 43 .The match between the XRD patterns of the C36 gels and that of the neat B O (orthorhombic) phase (Figure 3) is definitive.In addition, using optical microscopy methods that have been known for more than a century 46 , the orientation of the long molecular axes of individual molecules has been shown to be orthogonal to the planes defined by the microplatelets that constitute the building blocks of the supermolecular assembly.This is the first time a complete structural determination of the solid component of a gel has been made.

Figure 3 .
Figure 3. Overlay of the column graph representation of XRD data for crystalline C36 in its B O phase 45 and the XRD patterns for gels composed of C36 at (a) 4 wt% in 1-octanol, (b) 2 wt% in 1-octanol, (c) 4 wt% in hexadecane and (d) 4 wt% in glycidyl methacrylate (adapted from ref 43).

Table 1 .
Transition temperatures, T gel ( o C), and periods of stability (at room temperature) a of gels with nNBr gelators.
rate b (T gel , o C)