Regioselectivity of acid-catalyzed ring-opening of epoxides

Not to be confused with what is the mechanism of acid-catalyzed ring opening of epoxides.

What is the correct order of regioselectivity of acid-catalyzed ring-opening of epoxides: $3^\circ$ > $2^\circ$ > $1^\circ$ or $3^\circ$ > $1^\circ$ > $2^\circ$? I am getting ready to teach epoxide ring-opening reactions, and I noticed that my textbook has something different to say about the regioselectivity of acid-catalyzed ring-opening than what I learned. My textbook does not agree with 15 other introductory texts I own, but it does agree with one. None of my Advanced Organic Chemistry texts discuss this reaction at all. Thus, I have no ready references to go read.

Edit: My textbook is in the minority, and it is a first edition. Is it wrong, or do the other 15 texts (including some venerable ones) oversimplify the matter?

What I learned

Although the acid-catalyzed ring-opening of epoxides follows a mechanism with SN2 features (inversion of stereochemistry, no carbocation rearrangements), the mechanism is not strictly a SN2 mechanism. The transitition state has more progress toward the C-LG bond breaking than an SN2, but more progress toward the C-Nu bond forming than SN1. There is significantly more $\delta ^+$ character on the carbon than in SN2, but not as much as in SN1. The transition states of the three are compared below:

In a More O’Ferrall-Jencks diagram, the acid-catalyzed ring-opening of epoxides would follow a pathway between the idealized SN2 and SN1 pathways.

Because of the significant $\delta ^+$ character on the carbon, the reaction displays regioselectivity inspired by carbocation stability (even though the carbocation does not form): the nucleophile preferentially attacks at the more hindered position (or the position that would produce the more stable carbocation if one formed). If a choice between a primary and a secondary carbon is presented, the nucleophile preferentially attacks at the secondary position. If a choice between a primary and a tertiary carbon is presented, the nucleophile preferentially attacks at the tertiary position. If a choice between a secondary and a tertiary carbon is presented, the nucleophile preferentially attacks at the tertiary position.

The overall order of regioselectiveity is $3^\circ$ > $2^\circ$ > $1^\circ$.

What my textbook says

My text agrees that the mechanism is somewhere in between the SN2 and SN1 mechanisms, but goes on to say that because it is in between, electronic factors (SN1) do not always dominate. Steric factors (SN2) are also important. My text says that in the comparison between primary and secondary, primary wins for steric factors. In other words, the difference between the increased stabilization of the $\delta ^+$ on secondary positions over primary positions is not large enough to overcome the decreased steric access at secondary positions. For the comparison of primary and tertiary, tertiary wins. The increased electronic stabilization at the tertiary position is enough to overcome the decreased steric access at the tertiary position. The comparison between secondary and tertiary is not directly made, but since $3^\circ$ > $1^\circ$ and $1^\circ$ > $2^\circ$, it is implied that $3^\circ$ > $2^\circ$.

If this pattern is true, then other cyclic "onium" ions (like the bromonium ion and the mercurinium ion) should also behave this way. They don't.

Typical of introductory texts, no references are provided. A Google search did not yield satisfactory results and the Wikipedia article on epoxides is less than helpful. Since I 15 other introductory texts on my bookshelf, I consulted all of them on this reaction. The following is a summary of my findings. Only two of the texts (the one I am using and one other) describe the regioselectivity as $3^\circ$ > $1^\circ$ > $2^\circ$. All of the other books support the other pattern, including Morrison and Boyd (which lends credence to the pattern that I learned).

Books that have $3^\circ$ > $2^\circ$ > $1^\circ$

• Brown, Foote, Iverson, and Anslyn
• Hornsback
• Ege
• Bruice
• Smith
• Fessenden and Fessenden
• Volhardt and Schore
• Solomons and Fryhle
• Jones
• Baker and Engel
• Ouellette and Rawn
• Carey
• Morrison and Boyd
• Streightweiser and Heathcock

Books that have $3^\circ$ > $1^\circ$ > $2^\circ$

• Klein (the text I am using)
• McMurray

I also surveyed my various Advanced Organic texts (March, Smith, Carey and Sundberg, Wyatt and Warren, Lowry and Richardson, etc.). Interestingly, none of them even mention acid-catalyzed ring-opening of epoxides (either by Brønsted or Lewis acids). I suspect that these omissions mean that this reaction 1) has difficult to predict regioselectivity (despite the predominance of introductory books that suggest otherwise), and thus 2) is synthetically useless. If #2 is true, then why is this reaction in introductory organic texts?

• In the first equation (me-subst epoxide) there is an extra carbon in the minor product. (Same for the variation of it.) – Martin - マーチン Jun 18 '14 at 9:14
• Unless I'm missing something, the axis labels on the MOFJ plot are reversed. – orthocresol Sep 2 '18 at 7:55
• I have read this discussion several times and finally got motivated to do a Chemical Abstracts search on line for the reaction of propylene oxide and methanol under acid catalysis. Many of the catalysts were zeolites and rare earths. The overwhelming conclusion is that the epoxide opening with methanol is at the primary site leading to the secondary alcohol. – user55119 Jan 25 '19 at 13:30

First part

It won't decide the issue but the Organic Chemistry text by Clayden, Greeves, Warren and Wothers also mentions that the matter might not be as clear-cut as the majority of your textbooks make it seem. This might strengthen the position of the textbook you're using a bit. But again, there are no references given. Here is the relevant passage (especially the last two paragraphs):

Second Part

I have found the following passage on the formation of halohydrins from epoxides in the book by Smith and March (7th Edition), chapter 10-50, page 507:

Unsymmetrical epoxides are usually opened to give mixtures of regioisomers. In a typical reaction, the halogen is delivered to the less sterically hindered carbon of the epoxide. In the absence of this structural feature, and in the absence of a directing group, relatively equal mixtures of regioisomeric halohydrins are expected. The phenyl is such a group, and in 1-phenyl-2- alkyl epoxides reaction with $\ce{POCl3}/\ce{DMAP}$ ($\ce{DMAP}$ = 4-dimethylaminopyridine) leads to the chlorohydrin with the chlorine on the carbon bearing the phenyl.${}^{1231}$ When done in an ionic liquid with $\ce{Me3SiCl}$, styrene epoxide gives 2-chloro-2-phenylethanol.${}^{1232}$ The reaction of thionyl chloride and poly(vinylpyrrolidinone) converts epoxides to the corresponding 2-chloro-1-carbinol.${}^{1233}$ Bromine with a phenylhydrazine catalyst, however, converts epoxides to the 1-bromo-2-carbinol.${}^{1234}$ An alkenyl group also leads to a halohydrin with the halogen on the carbon bearing the $\ce{C=C}$ unit.${}^{1235}$ Epoxy carboxylic acids are another example. When $\ce{NaI}$ reacts at pH 4, the major regioisomer is the 2-iodo-3- hydroxy compound, but when $\ce{InCl3}$ is added, the major product is the 3-iodo-2-hydroxy carboxylic acid.${}^{1236}$

References:

${}^{1231}$ Sartillo-Piscil, F.; Quinero, L.; Villegas, C.; Santacruz-Juarez, E.; de Parrodi, C.A. Tetrahedron Lett. 2002, 43, 15.

${}^{1232}$ Xu, L.-W.; Li, L.; Xia, C.-G.; Zhao, P.-Q. Tetrahedron Lett. 2004, 45, 2435.

${}^{1233}$ Tamami, B.; Ghazi, I.; Mahdavi, H. Synth. Commun. 2002, 32, 3725.

${}^{1234}$ Sharghi, H.; Eskandari, M.M. Synthesis 2002, 1519.

${}^{1235}$ Ha, J.D.; Kim, S.Y.; Lee, S.J.; Kang, S.K.; Ahn, J.H.; Kim, S.S.; Choi, J.-K. Tetrahedron Lett. 2004, 45, 5969.

${}^{1236}$ Fringuelli, F.; Pizzo, F.; Vaccaro, L. J. Org. Chem. 2001, 66, 4719. Also see Concellón, J.M.; Bardales, E.; Concellón, C.; García-Granda, S.; Díaz, M.R. J. Org. Chem. 2004, 69, 6923.

• I have the fifth and sixth editions of March, presumably something similar is present in the earlier editions. I will check. In the meantime, posting links to those references anyway would be helpful to the rest of the community who may not have access to a copy. – Ben Norris Feb 13 '14 at 14:43
• @BenNorris Ok, I add them. By the way, for the 6th Edition of March you find the relevant passage on p. 583. – Philipp Feb 13 '14 at 14:49

Extending the pattern of epoxides to bromonium ions and mercurinium ions is hazardous at best. Below the second row (the carbon row) bond lengths get much longer, and acid strengths are much lower than compared to sulfuric acid in methanol.

In general acid-catalyzed reactions involving epoxides where selectivity is a concern will have carbocation intermediates, but the are not required to do such. But this potential is most likely to scare off a potential synth-jock simply because it is effectively a lower throughput at whatever step.

The pedagogical value of teaching epoxides at the undergraduate level is 4-fold:

1. It allows discussion of the physical chemistry of strained ring-systems
2. Synthetic Organic Chemistry is the business of making and breaking carbon-carbon bonds. Via Grignard reagent, epoxides allow a two-carbon extension of a carbon skeleton, which is fairly simple to understand and at the point where the topic is introduced, often the first example.
3. It can serve as an introduction to polymer chemistry.
4. In the example you've presented, it forces students to evaluate the relative strengths of competing phenomena, which is the sine qua non of organic chemistry as an educational tool. In this case to explicitly address the question: Does the effect 2o carbocation stabilization for gain outweigh the steric hindrance?

If the instructor or students are obsessed with the 'right' answer rather than an intelligently described answer (which may also be right), epoxides are less useful as a teaching topic.

• I agree with everything you say, but it does not answer the initial question. Which regioselectivity is correct? My textbook is in the minority, and it is a first edition. Is it wrong, or do the other 15 oversimplify the matter? I have edited my question to reflect this latest thought on the matter. – Ben Norris Mar 5 '13 at 22:57
• I do not believe the regioselectivity can be predicted reliably outside of an actual experiment. – Lighthart Aug 5 '14 at 4:01
• I agree with this answer. It is possible to find examples that support both textbooks. – Aaron Oct 2 '15 at 7:05
• A few thoughts about the regioselectivity issue. 1) Do you believe that all text authors consult the original literature? 2) Is there conflicting information in the the original literature? – user55119 Apr 15 '18 at 14:22