Diastereomers

Diastereomer chemistry is not a historical curiosity — it is the daily working reality of pharmaceutical synthesis, natural-product total synthesis, and structural biology. Drugs with multiple stereocenters, such as the HIV protease inhibitor saquinavir (four stereocenters) or the antibiotic erythromycin (eighteen stereocenters), are synthesized or fermented as single diastereomers because different diastereomers of the same connectivity can have wildly different, even opposite, biological activity — a diastereomer is not a weaker version of the drug, it is often pharmacologically inert or actively toxic, because a biological receptor is itself chiral and reads out three-dimensional shape with the same discriminating power an NMR spectrometer applies to bulk physical properties. This is precisely why regulatory agencies require full stereochemical characterization, not just constitutional structure, before drug approval. Diastereomeric salt formation remains a workhorse industrial technique for resolving racemic mixtures: treating a racemic acid with a single-enantiomer chiral amine (e.g., resolving racemic tartaric acid derivatives historically, or resolving pharmaceutical intermediates with cinchona alkaloids today) converts a pair of enantiomers into a pair of diastereomeric salts, which — unlike the original enantiomers — have different solubilities and can be separated by simple fractional crystallization, then the resolving agent is removed to recover single-enantiomer product. This trick, dating to Pasteur's original tartrate work and refined into the modern "classical resolution" technique, only works because diastereomers (unlike enantiomers) have different physical properties. In carbohydrate and nucleic acid chemistry, the entire logic of sugar nomenclature (D-glucose vs. D-galactose vs. D-mannose) is a diastereomer map: these compounds share the same molecular formula and connectivity, differing only in the configuration at one or more of several stereocenters, giving each a genuinely distinct biological identity — different enzymes recognize them, different lectins bind them, different metabolic pathways process them. In modern medicinal chemistry, diastereomer-resolved synthesis underlies the production of statins (atorvastatin), antibiotics (many β-lactams), and the ever-expanding class of oligonucleotide and peptide therapeutics, where stereochemically-controlled solid-phase synthesis must avoid epimerization at every step precisely because a single wrong-configuration diastereomer at one residue among dozens is a distinct molecule with distinct (often reduced) target affinity. Analytically, diastereomer formation is exploited deliberately in Mosher ester analysis, where a chiral alcohol of unknown absolute configuration is converted (via esterification with both enantiomers of Mosher's acid chloride) into a pair of diastereomeric esters whose distinct NMR shifts allow the original alcohol's absolute configuration to be assigned — turning the very fact that diastereomers are NMR-distinguishable into a routine structure-determination tool used in nearly every modern total-synthesis paper.

Ask the tutor: Can you walk through why cis- and trans-2-butene give different diastereomers of 2,3-butanediol upon OsO₄ dihydroxylation, using full curved arrows?

Start with a molecule that has exactly two stereocenters, $C_2$ and $C_3$, each independently able to be $R$ or $S$. There are four combinations: $(2R,3R)$, $(2S,3S)$, $(2R,3S)$, $(2S,3R)$. Pair them by full mirror inversion — flipping both centers simultaneously — and you get two enantiomeric pairs:

Now compare across the pairs: $(2R,3R)$ versus $(2R,3S)$. These share the same configuration at C2 but differ at C3 — only one of the two centers has been inverted, not both. They are not mirror images of each other (mirror inversion would have to flip every stereocenter), yet they are not identical either. This is a diastereomeric relationship. The same logic applies to $(2R,3R)$ vs.\ $(2S,3R)$, and to every other "some but not all centers flipped" comparison. This is worth stating as a clean rule, because it is the operational test used in every exam problem: given two structures with identical connectivity, look at every stereocenter (and every alkene geometry). If all are inverted relative to each other → enantiomers. If none are inverted (all match) → identical (same compound, possibly drawn differently). If some but not all are inverted → diastereomers. Physically, why does this matter so much? Enantiomers are congruent to each other under reflection — literally, if you built physical models, one is exactly the mirror image of the other, atom for atom, distance for distance. Every intramolecular distance (every H···H, H···OH, C···C nonbonded contact) in one enantiomer has an exact numerical twin in the other, because reflection preserves all distances. Consequently every scalar physical property — melting point, boiling point, density, IR frequencies, $^1$H/$^{13}$C NMR shifts in an achiral solvent — is identical between enantiomers. Diastereomers, in contrast, are not mirror images: the act of flipping only one center out of two changes real, measurable through-space distances between the unflipped and flipped substituents. A gauche interaction that existed in one diastereomer may become anti in the other; a 1,3-diaxial clash present in one chair-flipped diastereomer may be absent in the other. These are different molecules in the same sense that cis- and trans-2-butene are different molecules — indeed, cis/trans alkene isomers are themselves a limiting case of the diastereomer relationship, since they are stereoisomers, not mirror images, and not identical.

Now consider the mechanistic question: when a reaction creates a new stereocenter next to one that already exists, what decides which diastereomer forms? The molecule's existing stereocenter creates a chiral, non-planar environment, and the incoming reagent approaches from whichever face is less hindered or better orbital-aligned — this is diastereofacial selectivity, and it is why nucleophilic additions to α-chiral carbonyls (Felkin–Anh), catalytic dihydroxylation (syn-addition across a chiral-environment alkene), and aldol reactions (Zimmerman–Traxler chair transition states) all give predictable, often overwhelmingly selective, single diastereomers rather than statistical mixtures.

The stereoisomer count for a molecule with $n$ independent stereocenters (no internal mirror symmetry) is

where: $n$ is the number of stereocenters that are each configurationally independent (this bound is refined by the Counting Stereoisomers: 2^n Rule when meso forms collapse the count). These $2^n$ stereoisomers organize into enantiomeric pairs,

where: each pair consists of the molecule and its full mirror image (every stereocenter inverted simultaneously). The number of diastereomeric relationships — pairs of stereoisomers that are not mirror images and not identical — for $n=2$ follows directly from total pairwise comparisons minus the identical and enantiomeric ones:

where: $\binom{2^n}{2}$ counts all distinct unordered pairs among the $2^n$ stereoisomers, and $2^{n-1}$ of those pairs are the enantiomeric (mirror) ones, leaving the rest as diastereomeric. Reading this for $n = 2$: there are $2^2 = 4$ stereoisomers, $\binom{4}{2} = 6$ total pairs, of which $2^{2-1}=2$ pairs are enantiomeric, leaving $6-2=4$ diastereomeric pairs — matching exactly the four diagonal/vertical arrows drawn in the figure above. For diastereoselectivity in a reaction, the outcome is governed by the relative rates of formation of competing transition states leading to different diastereomers. If $\Delta\Delta G^{\ddagger}$ is the free-energy difference between the two diastereomeric transition states,

where: $\Delta\Delta G^{\ddagger} = G^{\ddagger}_{\text{minor}} - G^{\ddagger}_{\text{major}}$, $R$ is the gas constant, $T$ is absolute temperature — this is the same Curtin–Hammett-type reasoning used throughout stereoselective synthesis, and it is what "predicting the diastereomer" mechanistically actually means: identifying which transition-state geometry has lower strain/better orbital overlap.

Example 1 — Foundational: assigning the diastereomer relationship for tartaric acid's three stereoisomers. Tartaric acid, HOOC–CHOH–CHOH–COOH, has two stereocenters, C2 and C3, each bearing OH, H, COOH, and CH(OH)COOH. Because C2 and C3 have "mirror-symmetric" substituent sets (swap the two ends of the molecule and C2 becomes C3), this is a special case with an internal mirror plane possibility — foreshadowing Meso Compounds — but the diastereomer logic is identical to the general case. The four raw stereo-combinations are $(2R,3R)$, $(2S,3S)$, $(2R,3S)$, $(2S,3R)$. Step 1: Check $(2R,3S)$ against $(2S,3R)$ for whether they are actually distinct. Because the molecule has a $C_2$-type internal symmetry (rotating the whole molecule 180° swaps C2 and C3 labels while swapping R↔S under the priority reassignment), $(2R,3S)$ and $(2S,3R)$ describe the same molecule — this is the meso compound, achiral despite having two stereocenters, because it possesses an internal mirror plane relating its two halves. Step 2: So there are only three actual stereoisomers of tartaric acid: $(2R,3R)$, $(2S,3S)$, and meso-$(2R,3S)$. Step 3: Classify the pairs.

Step 4: Predict physical consequences. $(2R,3R)$- and $(2S,3S)$-tartaric acid are enantiomers: identical melting points (168–170 °C), identical solubility, equal-and-opposite $[\alpha]_D$ ($+12°$ and $−12°$ in water, a commonly tabulated literature value). Meso-tartaric acid, being a diastereomer of both, has a genuinely different melting point (140 °C), is optically inactive ($[\alpha]_D = 0$, from its own internal mirror-plane cancellation — not from being a racemic mixture), and is separable from the chiral pair by ordinary crystallization. This is a clean limiting-case check: as the substituent asymmetry that makes C2 ≠ C3 environments is removed (i.e., in the fully symmetric tartaric acid skeleton), one diastereomeric "pair" collapses to a single achiral meso form — exactly the boundary case the $2^n$ counting rule must handle carefully.

Example 2 — Applied: predicting the major diastereomer in an aldol reaction via the Zimmerman–Traxler transition state. Consider the lithium enolate of propiophenone-type ketone (an ethyl ketone enolate, $Z$-configured under standard kinetic LDA deprotonation conditions) reacting with benzaldehyde (PhCHO) to form a new C–C bond with two new stereocenters (Cα and the carbinol carbon). Step 1 — Establish enolate geometry. Kinetic deprotonation of an ethyl ketone with LDA at −78 °C, via a six-membered Ireland-model transition state that places the ketone's R group pseudo-equatorial, gives predominantly the $Z$-enolate (the OLi and the α-methyl cis across the C=C). Step 2 — Build the Zimmerman–Traxler chair. The aldol transition state is modeled as a six-membered ring: Li coordinates both the enolate oxygen and the incoming aldehyde oxygen, organizing a chair-like array in which the aldehyde's R group (here, phenyl) prefers the equatorial position to avoid 1,3-diaxial strain. Step 3 — Trace the curved arrows. Enolate α-carbon (nucleophilic, from the C=C–OLi resonance form with negative character on carbon) attacks the aldehyde carbonyl carbon; the C=O π electrons of the aldehyde flow onto oxygen, which is captured by Li⁺ in the same concerted, cyclic, chair transition state — a single pericyclic-like six-membered arrow-pushing pattern. Step 4 — Read the relative stereochemistry off the chair. With the $Z$-enolate methyl group forced into a pseudo-equatorial-adjacent arrangement by the ring geometry, and the aldehyde phenyl equatorial, the chair transition state places the new Cα–Me and the new carbinol C–OH in a specific relative orientation that collapses, upon ring-opening, to the syn (or "erythro") aldol diastereomer as the major product — this is the well-established outcome that $Z$-enolates give predominantly syn aldols via Zimmerman–Traxler chairs, while $E$-enolates give predominantly anti aldols. Step 5 — Diastereomer bookkeeping. The syn and anti aldol products have identical connectivity (same atoms bonded to same atoms) and differ only in the relative configuration at the two new stereocenters — precisely the diastereomer relationship. They have different $R_f$ values on TLC, different $^1$H NMR coupling constants between the Cα–H and the carbinol C–H (typically $J_{syn} \approx 4$–6 Hz vs. $J_{anti}\approx 7$–9 Hz in six-membered cyclic derivatives or acyclic analysis via Newman projections), and are cleanly separable — the diastereomer relationship is what makes the $J$-coupling analysis diagnostic in the first place.

Example 3 — Exam/competition level: counting and classifying stereoisomers of a trisubstituted cyclohexane with a plane of potential symmetry. Consider 1,2,4-trimethylcyclohexane (three stereocenters at C1, C2, C4, each bearing a methyl group; note C1 and C4 are not related by the kind of internal symmetry that produces a meso form here because the substitution pattern is not symmetric about the ring the way 1,2- or 1,3-disubstitution can be — this problem intentionally uses an asymmetric numbering to avoid a hidden meso trap). Step 1 — Count naive stereoisomers. With $n = 3$ independent stereocenters and no molecular symmetry relating any two of them (1, 2, and 4 positions are constitutionally distinguishable from each other around the ring — position 2 is adjacent to 1 but not to 4, so no operation maps the substituent set onto itself), the full count is

Step 2 — Organize into enantiomeric pairs.

Explicitly: label configurations as a triple (C1, C2, C4) each R or S. The four pairs are

Step 3 — Pick two stereoisomers from different pairs and confirm the diastereomer relationship mechanistically (i.e., by direct comparison, not by rule-recitation). Compare $(R,R,R)$ to $(R,R,S)$: C1 matches (R=R), C2 matches (R=R), C4 differs (R vs. S). Some but not all centers differ → diastereomers. Physically, this means the two compounds have different relative orientations of the C4-methyl group relative to the C1/C2 methyls (one places all three methyls on a "matching" cis/trans-relationship pattern around the ring; flipping just C4 converts, say, a 1,2-cis-4-cis arrangement into a 1,2-cis-4-trans arrangement) — a genuinely different ring-substitution topology with different 1,3-diaxial interactions and therefore a different conformational energy and physical properties. Step 4 — Limiting-case sanity check. If instead the molecule had a genuine internal symmetry (say 1,3,5-trimethylcyclohexane, where C1, C3, C5 are related by the ring's 3-fold pattern under certain substitution combinations), some of the eight naive stereoisomers would coincide, reducing the count below $2^n=8$ — exactly the correction the Counting Stereoisomers: 2^n Rule formalizes for symmetric substrates. For the deliberately asymmetric 1,2,4-numbering used here, no such collapse occurs, and the full count of 8 stands, giving 4 enantiomeric pairs and $\binom{8}{2} - 4 = 28-4 = 24$ diastereomeric pair-relationships among them.

The conceptual foundation for distinguishing diastereomers from enantiomers rests on the 19th-century stereochemical revolution associated above all with Jacobus Henricus van 't Hoff and Joseph Achille Le Bel, who independently proposed in 1874 that the four bonds of carbon are directed tetrahedrally in space — the geometric fact that makes stereoisomerism possible at all. Van 't Hoff went further, in his 1875 pamphlet Voorstel tot Uitbreiding der Tegenwoordige Chemische Structuurtheorieën, to explicitly work out that a compound with $n$ asymmetric carbons could have up to $2^n$ stereoisomers, and to recognize that not all pairs among these stereoisomers are mirror images — laying the combinatorial and physical groundwork for the enantiomer/diastereomer distinction articulated more fully once Emil Fischer's sugar work (the 1890s determination of the configurations of the aldohexoses) provided a concrete, richly-populated worked example: glucose, galactose, mannose, and the rest of the sixteen aldohexose stereoisomers are a textbook-perfect diastereomer/enantiomer network, eight D-sugars each paired with its L-enantiomer, and every D-sugar a diastereomer of every other D-sugar.