The gate of Q1 is actively driven high by a 500ns pulse from my trusty PIC, shown here in YELLOW. In a real circuit the drain of Q1 would be the output to the next circuit, depicted in RED. Q2, in concert with R1 and R2 (4.7K), simulate the load FET in the 4004.
In the next circuit I've disconnected Q2's gate from Vdd and put another FET in the path. This FET, Q3, has its gate connected to Vdd, acting as another resistor. Unlike the previous circuit there won't be significant current flow if this FET turns on fully, so I haven't bothered with additional resistors.
Next, I added a 33pF bootstrap capacitor C1 between Q2's gate and the "source" of our simulated load FET. As I explained in Puzzling out the Bootstrap Load, Q3 will act as a diode, allowing the bootstrap capacitor to charge when the load FET's source is pulled low, but preventing it from discharging when the source is allowed to rise toward Vdd.
Just for fun I went one step further and replaced Q3 with a 1N914 fast switching diode. As you can see below, this improves the charge pump efficiency over using the BSS83 MOSFET with Q2's Vgs reaching 3V. Note the vertical scales are are 2V/div in this trace, rather than to 1V/div in the previous traces.