The PHICH channel is used to transmit hybrid-ARQ (HARQ) acknowledgements in response
to uplink shared channel (UL-SCH) transmission from terminals.
The acknowledgements are represented by 1-bit data where 0 indicates NACK and 1 indicates ACK. If an ACK is incorrectly decoded as NACK by the terminal, this leads to an unnecessary uplink retransmission of a correctly decoded transport block. If a NACK is incorrectly decoded as ACK by the terminal, this leads the loss of a transport block. So the error rate requirement of PHICH is stricter than PCFICH . To meet such requirement, the PHICH is heavily encoded such as using BPSK instead of QPSK and multiple PHICHs are mapped into a PHICH group so each PHICH is separated into several REGs. Typically, the PHICH is mapped onto the first OFDM symbol so the terminal will always decode the PHICH even if the PCFICH decoding failed.
So answer to your question why the modulation will come first because to reduce the unnecessary re-transmissions we are sending 3-bits together. It would be always same either 000 or 111. So first it will repeat the 1-bit to 3-bits (Modulation) and then sampling after that again modulation will happen depends upon the band using (QPSK or BPSK). Which will build a group called PHICH group. Further Resource group alignment, layer mapping, precoding, maping PHICH to Resource element. So this is how it works differently from other channels.