CTS frame contains duration for sending of a node which requests it. And after the node sends the whole frame without collision, the AP broadcasts ACK frame. Then when does another node sends its RTS frame between "after the duration" or "after it receives ACK" ?
The method of channel access you refer to that is based on the duration field in IEEE802.11 frames uses the concept of the network allocation vector (NAV).
Each station maintains such a NAV and updates it after receiving any frame that contains a duration entry. The sender of the frame indicates to the other parties that share the same Wi-Fi channel that they should back-off for the time indicated in the duration field. The value in the NAV is updated and increased, when the value in the duration field is larger than the current value inside the NAV. After the time in the NAV is expired and no further update to the NAV was received, the station will be free to access the channel via this virtual carrier sensing mechanism.
So in your practical example, the listening station sets its NAV according to the duration value of the CTS-frame, which projects to the time, when the ACK frame is finished. When the data is not fragmented to several frames or in the case of the last frame of data fragmented to several frames, the ACK frame will include a duration value of 0, which leads to no update of the NAV. Therefore, the station will not wait for an ACK frame to access the medium again.
The ACK frame has an influence on the NAV is in the case of data fragmented to several frames. Here the ACK frames following the intermediate data frames will carry a finite duration value which indicates to the station that it has to extend the NAV, as more data frames will be following.