Postgresql indexes not used - postgresql

Question

I'm trying to understand why indexes are not used by this query:
SELECT
te.idturno
FROM turno t
JOIN turnoestado te ON t.idturno = te.idturno
ORDER BY te.idturno, te.idturnoestado
Limit 100
Row count for table Turno is ~1 million, and TurnoEstado is ~10 million.
Here are the index definition:
CREATE INDEX idx_turnoestado_idturno
ON turnoestado USING btree (idturno);
CREATE INDEX idx_turnoestado_idturnoestado
ON turnoestado USING btree (idturnoestado);
As you can see, there are two indexes for the keys used in the Order By clause, but none of them are used acording to the query plan returned by explain analyze:
Limit (cost=988700.56..988700.81 rows=100 width=8) (actual time=7332.740..7332.754 rows=100 loops=1)
-> Sort (cost=988700.56..1014125.36 rows=10169918 width=8) (actual time=7332.737..7332.746 rows=100 loops=1)
Sort Key: te.idturno, te.idturnoestado
Sort Method: top-N heapsort Memory: 29kB
-> Hash Join (cost=160603.86..600013.61 rows=10169918 width=8) (actual time=505.221..6279.233 rows=10169918 loops=1)
Hash Cond: (te.idturno = t.idturno)
-> Seq Scan on turnoestado te (cost=0.00..178003.18 rows=10169918 width=8) (actual time=0.011..1249.563 rows=10169918 loops=1)
-> Hash (cost=143894.94..143894.94 rows=1018394 width=4) (actual time=504.731..504.731 rows=1018394 loops=1)
Buckets: 4096 Batches: 64 Memory Usage: 571kB
-> Seq Scan on turno t (cost=0.00..143894.94 rows=1018394 width=4) (actual time=0.033..342.071 rows=1018394 loops=1)
Total runtime: 7332.824 ms
If I use only one key in the Order By, the result is instantaneous, the same happens if I create a multicolumn index. Why the planner does not use the indexes already created?.
P.S.: I'm using "PostgreSQL 9.3.5 on x86_64-unknown-linux-gnu, compiled by gcc (Ubuntu 4.8.2-19ubuntu1) 4.8.2, 64-bit"