Read on Twitter
here's a cross section of the MAX233 RS-232 driver and receiver. this part doesn't need external capacitors--because there are four of them built right in!
think that's amazing? check out the underside! this is the complex custom lead frame that connects those four capacitors to the chip.
here's how it lines up with the pinout. (i've flipped the lead frame pic). the datasheet wants you to connect some pins together because this wires up some of the capacitors!
let's look at some interesting details. these scalloped patterns are where i've sliced into the conductive plates. the cut is not parallel to the plates because the capacitors are *soldered* to the lead frame and are slightly crooked!
here's a capacitor again up close. you can see alternating layers of conductor and ceramic insulator. also the copper plating on the end caps is readily apparent.
zooming in on the silicon chip, see those little gold dots around the outside? those are the ends of the bond wires. they attach to the metal lead frame which you can't see from this angle. then they arc up and over onto the chip. you can see the gold dots there too.
this is what the silicon chip looks like under my crappy microscope. it is a bit scarred up because i removed the plastic using a scalpel. there are some big transistors that form part of the charge pump circuit...
...this charge pump circuit takes the 5V in, charges up the flying capacitor, and the connects it "on top of" the 5V supply so you get 10V out.
another charge pump circuit takes the 10V, charges another flying capacitor, and then *swaps the polarity of it* in order to get -10V. these voltages provide the logic high and low outputs for the RS232 pins.
if i really push my microscope to the limit, you can see the blurry mess that is the official die ID: "RS30". this is the *same* die used in the MAX232!
another fun pic of one of the embedded surface mount capacitors in the MAX233.
the chip is very thin now 
