Take 4-hydroxybutan-2-one, HOCH₂CH₂COCH₃. It has a ketone at one end and an alcohol at the other. Suppose you want CH₃MgBr to add a methyl group to the ketone. The Grignard reagent never gets there. It reacts with the alcohol’s O–H first and is used up. Other carbon nucleophiles that add to ketones, such as CH₃Li, are at least as basic, so switching reagents does not help.
This is the general problem. Every reagent in this course has been described by what it does to one functional group, but real molecules often carry several. A reagent that cannot tell two of them apart attacks both. When no choice of reagent solves that, the way out is to hide one group for a while.
A protecting group is a temporary mask: you attach it to a functional group so that group cannot react, and remove it later. It must go on cleanly, survive the step you actually want, and then come off under conditions that leave the rest of the molecule alone. Acetals & hemiacetals set out these same three requirements. The cost is three steps (protect, react, deprotect) where you wanted one, so you use a protecting group only when nothing simpler works.
The classic problem: a Grignard and an alcohol in one molecule
A Grignard reagent behaves as a carbanion, and a carbanion is a very strong base. An alcohol’s O–H has a pKa near 16. The alkane the carbanion would become has a pKa near 50, so the O–H is vastly more acidic, and the Grignard pulls its proton off at once:
RMgBr + R'OH → R–H + R'OMgBr
The Grignard has become an alkane before it can reach any carbonyl. This is not a slow side reaction. Proton transfer is the fastest reaction open to it. An N–H, an S–H or a carboxylic acid’s O–H destroys a Grignard in the same way, and so does water, which is why Grignard reactions are run in thoroughly dry ether. Why C–metal means nucleophilic carbon ranks these acids by pKa.
Protecting an alcohol: a silyl ether
Return to 4-hydroxybutan-2-one. Treat it with tert-butyldimethylsilyl chloride (TBSCl) and imidazole, a mild nitrogen base that takes up the HCl released. The O–H becomes an O–Si bond. The product is a silyl ether: an ether in which one of the two groups on oxygen is a silicon group instead of a carbon group. The silicon group here, Si(CH₃)₂C(CH₃)₃, is called TBS. A silyl ether has no acidic proton left, so a Grignard ignores it.
The silyl ether comes off with fluoride, usually as TBAF (tetrabutylammonium fluoride, Bu₄N⁺ F⁻, a fluoride salt that dissolves in organic solvents). The Si–F bond is exceptionally strong, so fluoride pulls the silicon off the oxygen. It needs no strong acid and no oxidant, so almost everything else in the molecule is left alone.
The bulky tert-butyl group on silicon matters. It crowds the silicon, and that crowding lets a TBS ether survive a quick, mild aqueous workup. A small trimethylsilyl ether, R–O–Si(CH₃)₃, would fall apart under the same conditions. TBS is not acid-proof, though. Longer exposure to acid takes it off, and acetic acid in aqueous THF, or HCl in methanol, is a standard way to remove it.
Protecting a ketone or aldehyde: an acetal
A ketone or aldehyde is protected as a cyclic acetal, which Acetals & hemiacetals builds step by step. Treat the carbonyl compound with ethylene glycol, HOCH₂CH₂OH, and an acid catalyst while removing water. The two oxygens of the diol replace the C=O, and a five-membered ring containing two oxygens forms.
The acetal has no C=O and no electrophilic carbon, so hydride reagents and Grignard reagents have nothing to attack. It comes off with aqueous acid. Acetal formation is an equilibrium: removing water drives it forward, and adding water back with acid runs the same reaction in reverse. That is why the deprotection is so reliable.
Neither mask changes an oxidation level. Count the bonds from the carbonyl carbon to oxygen: two before protection (the C=O) and two after (two C–O single bonds). In a silyl ether, the carbon keeps its single bond to oxygen. So protecting and deprotecting costs steps but never oxidizes or reduces anything.
An acetal can also tell two carbonyls apart. Give a molecule that holds both an aldehyde and a ketone one equivalent of the diol, and only the aldehyde is protected, because it is the more reactive carbonyl: more electrophilic and less hindered (Acetals & hemiacetals explains why). With the aldehyde masked, a route can reduce the ketone or add a Grignard to it. Without the mask, no reagent in this course could do that, because any reagent that attacks a ketone attacks the more reactive aldehyde first.
Protecting an amine: a carbamate
Take butan-1-amine, CH₃CH₂CH₂CH₂NH₂. Its nitrogen lone pair makes it both a nucleophile and a base. It attacks acyl chlorides and other acylating reagents, and alkyl halides too.
The usual mask is a carbamate: a carbonyl carbon bonded to both a nitrogen and an OR group, R–NH–C(=O)–OR′. Think of it as an amide and an ester sharing one carbonyl. As in an amide (Esters & amides), the nitrogen lone pair is shared with the C=O by resonance. So the nitrogen becomes a very poor nucleophile, and a very weak base.
This carbamate still has one N–H, though, so it does not make the molecule safe for a Grignard. Its job is to stop the nitrogen from attacking.
Two carbamates are standard, and they differ only in the OR′ group. Boc (tert-butoxycarbonyl) carries O–C(CH₃)₃. It goes on with Boc₂O (di-tert-butyl dicarbonate), an anhydride-like reagent whose carbonyl the amine attacks, just as it attacks an acid anhydride in Acid chlorides & anhydrides. Cbz (benzyloxycarbonyl) carries O–CH₂–C₆H₅. It goes on with benzyl chloroformate, C₆H₅CH₂OCOCl, which reacts like an acyl chloride.
Boc comes off with a strong acid, usually TFA (trifluoroacetic acid, CF₃COOH). TFA protonates the carbonyl oxygen. That makes the neighboring O–C(CH₃)₃ bond break easily, and the tert-butyl group leaves as a tertiary carbocation, which then loses a proton to become 2-methylpropene. What remains is a carbamic acid, R–NH–COOH. A carbamic acid loses CO₂ on its own and gives back the amine. In the acid solution the amine is protonated, so a basic workup is needed to free it.
Cbz comes off with H₂ and Pd/C instead. The bond between the oxygen and the benzylic CH₂ breaks, and a hydrogen caps each end. Breaking a single bond with H₂ over a metal catalyst like this is called hydrogenolysis. It is the same reaction that removes a benzyl ether in Benzylic reactivity. The products are toluene and the same carbamic acid, which again loses CO₂ to give the amine.
Protecting a carboxylic acid: a methyl ester
A carboxylic acid’s O–H, with a pKa near 5, gives up its proton to any base. The simple mask is a methyl ester, made by heating the acid with methanol and an acid catalyst. At the end, aqueous NaOH hydrolyzes the ester, and H₃O⁺ then gives back the acid. A methyl ester has no acidic proton, so it stops the acid–base problem. It is still a carbonyl, though, and Grignard reagents and LiAlH₄ attack esters, so this mask does not protect against them.
The four masks at a glance
| Group | Protect with | Becomes | Remove with |
|---|---|---|---|
| Alcohol | TBSCl, imidazole | Silyl ether | TBAF (F⁻) |
| Ketone / aldehyde | HOCH₂CH₂OH, H⁺ | Cyclic acetal | H₃O⁺ |
| Amine | Boc₂O, or benzyl chloroformate | Carbamate (Boc or Cbz) | Boc: TFA · Cbz: H₂, Pd/C |
| Carboxylic acid | CH₃OH, H⁺ | Methyl ester | NaOH, H₂O, then H₃O⁺ |
Orthogonality
Picture a molecule with two amines, one masked as Boc and one as Cbz. TFA removes the Boc and leaves the Cbz in place. H₂ over Pd/C removes the Cbz and leaves the Boc in place. Either amine can be freed first, while the other stays masked.
Protecting groups like these are called orthogonal: each one comes off under conditions the other survives. An amino acid is a small molecule that carries both an amine and a carboxylic acid. A peptide is a chain of amino acids joined by amide bonds, and chemists who build one rely on pairs like this. They build the chain one amino acid at a time, unmasking one amine at each step while the other masks stay on. Biomolecules covers peptides.
Not every pair is that clean. Fluoride removes a TBS ether and leaves an acetal untouched, because fluoride has nothing to react with in an acetal. The reverse is less reliable. Acid strong enough to remove the acetal can also remove a TBS ether, as the alcohol section showed. So with both masks on one molecule, take the TBS ether off first. If the acetal has to go first, the acid must be as mild and brief as possible, and even then some silyl ether may be lost.
Choosing a protecting group means checking it against every later step, not only the one you are protecting against. An acetal comes off in aqueous acid, so it cannot be carried through a later step that uses aqueous acid for some other reason.
Target: add a methyl group to the ketone of 4-hydroxybutan-2-one, HOCH₂CH₂COCH₃, to make a tertiary alcohol, without touching the existing OH.
The problem. CH₃MgBr would be destroyed by the free O–H before it reached the ketone.
1. Protect. TBSCl, imidazole. The alcohol becomes a TBS ether with no acidic proton.
2. React. CH₃MgBr, then a mild aqueous workup. The Grignard now survives long enough to attack the ketone. It gives a tertiary alkoxide, which the workup protonates to a tertiary alcohol. The TBS ether survives the brief workup.
3. Deprotect. TBAF. Fluoride removes the silicon and gives back the original OH. The new tertiary alcohol is untouched. The product is 3-methylbutane-1,3-diol.
The cost is three steps for one new bond. It is worth paying, because one equivalent of CH₃MgBr on the unprotected alcohol gives no product at all.
Target: 6-hydroxy-6-phenylhexan-2-one, PhCH(OH)CH₂CH₂CH₂COCH₃. From: 5-bromopentan-2-one, BrCH₂CH₂CH₂COCH₃, and benzaldehyde, PhCHO. Ph stands for a phenyl group, C₆H₅.
Read the target. Five chain carbons come from the bromoketone. The sixth, which carries the phenyl and the OH, comes from benzaldehyde. So the route makes one new C–C bond, between the benzaldehyde carbon and the carbon that carried the bromine. The OH sits where that bond forms, and the ketone stays where it was. That is a Grignard disconnection: the bromide becomes the nucleophile, and the aldehyde is the electrophile.
Find the conflict. You cannot make a Grignard from 5-bromopentan-2-one. As soon as one molecule of the reagent forms, it attacks the ketone of the next molecule, so the reagent eats its own starting material. The ketone is not a bystander here. It is the reason the obvious route fails.
1. Protect. Ethylene glycol with TsOH (p-toluenesulfonic acid, a strong acid that dissolves in organic solvents), removing water. The ketone becomes a cyclic acetal, 2-(3-bromopropyl)-2-methyl-1,3-dioxolane. The bromide is untouched, because a primary alkyl bromide does not react with a diol and a trace of acid.
2. Make the reagent. Mg, dry ether. The C–Br becomes C–MgBr. Nothing electrophilic is left in the molecule for the new Grignard to attack.
3. React. PhCHO. The Grignard adds to the aldehyde and gives a secondary alkoxide.
4. Deprotect. H₃O⁺. The acid protonates the alkoxide to the secondary alcohol, and it also hydrolyzes the acetal back to the ketone. The target is complete.
Notice that one aqueous acid workup does two jobs here, so the deprotection costs no extra step. Look for that saving whenever an acetal is the mask.
When not to use one
Protecting groups add two steps and lose some material in each step, so they are a last resort. Before reaching for one, ask three questions.
- Is there a milder reagent that is already selective? A molecule with a ketone and an ester needs no protection to reduce only the ketone, because NaBH₄ does it in one step. A protecting group here would mean three steps where one works.
- Can the steps be reordered? Often the interfering group can be installed after the sensitive step. That removes the problem entirely and beats protecting.
- Is the group genuinely reactive under these conditions? An ester or a ketone survives catalytic hydrogenation at one atmosphere, and a benzene ring survives most things. Not every functional group present is a threat.
What carries forward
Protecting groups are where synthesis stops being a list of reactions and becomes planning. The question is no longer which reaction to use, but in what order, and what has to be hidden while it happens. That is the subject of Planning a multistep route, the last section of this chapter.