The Romans used several types of locks for doors: the Laconian or primitive wooden pin tumbler lock, the padlock and the built-in or fastened-on metal lock. There are no locks in this collection that I can definitely identify as door locks. However, it seems that some of the larger barrel locks were used on doors, replacing the more primitive wooden locks. For example, the formidable 500 page volume on Roman antiquities in the British Museum (Guide to the Exhibition) has this to say about them: "Padlocks of Roman date are common. In this Case three of a barrel form are shown. One (No.400; fig. 173. has the key still rusted in it. The padlock has traces of a chain attachment at one end, and was probably kept hanging to a door post, while the bolt was shot into the end link of a chain attached to the door." Of course, such an arrangement would not allow the door to be locked from the inside, so bolts were used. These were placed at the bottom of each door, and fastened into the sill. (Smith's Dictionary). A bolt of iron or wood may also have been used.
There is another reason why door locks are not often found, aside from a few Byzantine examples such as that from Hoyuk with very substantial cases. There are many small escutcheons simply because everyone recognizes a keyhole, but no accompanying mechanisms. As Paul Barford (Paul-Barford.blogspot.com) points out, the finders of metal artifacts simply do not recognize the pieces they find as parts of rotary mechanism locks. There is nothing so obvious as a pin tumbler bolt or lock plate, and the pieces found seem to be just more Roman junk, of which considerable quantities turn up. I am afraid that most bronze door locks end up in the scrap bucket, with no hope of being reassembled. Iron components fare even worse.
As evidence of the metal door lock I first show, here the two illustrations by Stender
Figures 3 and 5 below)(archeologie-Krefeld.de/Bilder). These are images taken from a Roman sarcophagus, showing a model of a metal pin tumbler lock. This is now in the Deutsche Schloss-und Beschlagemuseum at Velbert. However, since a key could not be removed from an open pin tumbler lock, they were not very suitable as door locks, and seem not to have been used for this purpose. They were better adapted for containers: boxes, caskets and chests, that were not to be left open. I do not know why this carving was made for a sarcophagus, but it's doubtful that it refers to an actual use in a door lock.
Door lock, Pompeii This is the image of a door lock with a mechanism for a rotary key. This was found at Pompeii in the so-called "House of the Locksmith" (Towne) attached to the wood plank and with the key in place.
Lullingstone Barrel Lock Next is a nearly complete barrel lock found at Lullingstone Roman villa. (flikr/finkangel). It has a strip of metal attached that was obviously meant for mounting on a wood surface. I believe this to be a door lock. It would require the "bolt" to be drawn completely out of the lock when released by the key. The lock body itself would be fastened to the surface.
Hoyuk A and B Front and back of a Byzantine surface lock, thought from context to be from a stable (300-1100 CE) (Gorny)
Quennell ring key and lock. I'm repeating this sketch because it is a rare image of the operation of a lock for a rotary key. It could have been used either for a door or a chest.
Door Lock, turning key Compare this functional diagram by Raine (historicallocks) to that of Quennell. I think they are both showing the same basic mechanism, slightly different interpretation. However, I have never seen any surviving bolts that would serve in such a device. There must have been a very large number of them.
Door knob 5357 This was alleged to be a door knob and it may very well be. I have not seen one like it, and don't know how it might have been mounted. Nevertheless, it's a handsome artifact, and I will give it the benefit of the doubt until I learn more.