REGR: query raising for all NaT in object column

doc/source/whatsnew/v2.2.1.rst

@@ -27,6 +27,7 @@ Fixed regressions
 - Fixed regression in :meth:`DataFrame.loc` which was unnecessarily throwing "incompatible dtype warning" when expanding with partial row indexer and multiple columns (see `PDEP6 <https://pandas.pydata.org/pdeps/0006-ban-upcasting.html>`_) (:issue:`56503`)
 - Fixed regression in :meth:`DataFrame.map` with ``na_action="ignore"`` not being respected for NumPy nullable and :class:`ArrowDtypes` (:issue:`57316`)
 - Fixed regression in :meth:`DataFrame.merge` raising ``ValueError`` for certain types of 3rd-party extension arrays (:issue:`57316`)
+- Fixed regression in :meth:`DataFrame.query` with all ``NaT`` column with object dtype (:issue:`57068`)
 - Fixed regression in :meth:`DataFrame.shift` raising ``AssertionError`` for ``axis=1`` and empty :class:`DataFrame` (:issue:`57301`)
 - Fixed regression in :meth:`DataFrame.sort_index` not producing a stable sort for a index with duplicates (:issue:`57151`)
 - Fixed regression in :meth:`DataFrame.to_dict` with ``orient='list'`` and datetime or timedelta types returning integers (:issue:`54824`)

pandas/core/generic.py

@@ -597,7 +597,7 @@ def _get_cleaned_column_resolvers(self) -> dict[Hashable, Series]:
 
         return {
             clean_column_name(k): Series(
-                v, copy=False, index=self.index, name=k
+                v, copy=False, index=self.index, name=k, dtype=self.dtypes[k]
             ).__finalize__(self)
             for k, v in zip(self.columns, self._iter_column_arrays())
             if not isinstance(k, int)

pandas/tests/frame/test_query_eval.py

@@ -1415,3 +1415,11 @@ def test_query_ea_equality_comparison(self, dtype, engine):
             }
         )
         tm.assert_frame_equal(result, expected)
+
+    def test_all_nat_in_object(self):
+        # GH#57068
+        now = pd.Timestamp.now("UTC")  # noqa: F841
+        df = DataFrame({"a": pd.to_datetime([None, None], utc=True)}, dtype=object)
+        result = df.query("a > @now")
+        expected = DataFrame({"a": []}, dtype=object)
+        tm.assert_frame_equal(result, expected)