Method for improving yield rate using redundant wire insertion

1. A method for manufacturing an integrated circuit (IC) capable of tolerating random manufacturing defects, the method comprising:

generating, by a graphical construction unit, a first graph corresponding to a first net of the IC, the first graph representing a pin of the first net as a vertex, and a connection between two pins of the first net as an edge, the first graph further corresponding to a first IC layout;

identifying a first and a second pair of unconnected vertices in the first graph for inserting a first and a second redundant edge, respectively, the first redundant edge and the second redundant edge forming a first connected loop and a second connected loop, respectively, each connected loop further including at least two edges of the first graph;

calculating a tolerance ratio for the first redundant edge and the second redundant edge;

sorting the first and second redundant edge based on their tolerance ratio;

calculating a yield rate change of the first IC layout associated with inserting one of the first or second redundant edge with a highest tolerance ratio, and updating the first IC layout to include the redundant edge with the highest tolerance ratio if it is feasible and if the yield rate change is greater than zero; and

calculating the yield rate change of the first IC layout associated with inserting the first or second redundant edge having a second highest tolerance ratio, and updating the first IC layout to include the redundant edge with the second highest tolerance ratio if it is feasible and if the yield rate change is greater than zero.

2. The method of claim 1 , wherein generating the first graph further comprises:

calculating at least one projection point corresponding to a first vertex of the first net, the at least one projection point being calculated as a shortest distance from the first vertex to at least one neighboring edge; and

generating a first projection graph by modifying the first graph to include an additional vertex corresponding to the at least one projection point.

3. The method of claim 1 , wherein identifying the first and second pair of vertices further comprises:

identifying a different first pair of vertices if the second connected loop includes all of the at least two edges of the first connected loop, or a different second pair of vertices if the first connected loop includes all of the at least two edges of the second connected loop.

4. The method of claim 1 , wherein calculating the tolerance ratio comprises calculating the tolerance ratio of a redundant edge as

C

r



r



.

,

wherein |r| is a Manhattan length of the redundant edge, and C r is the total wire length of the connected and closed path formed by the redundant edge including the Manhattan length of the redundant edge.

5. The method of claim 1 , wherein calculating the yield rate change comprises calculating the yield rate change associated with a redundant edge as ΔY≅(C r −|r|)·o−ΔP′·s. wherein ΔY is the yield rate change associated with the redundant edge, |r| is a length of the feasible path corresponding to the redundant edge, C r is an increase in length of a feasible path corresponding to a connected path formed by the redundant edge, o is an open rate per unit wire length associated with the IC, ΔP′ is a change in parallel run length associated with inserting the redundant edge, and s is a short rate per unit parallel run length of wire in the IC.

6. The method of claim 1 , wherein selecting at least a first and second redundant edge in the first projection graph comprises selecting a plurality of first and second redundant edges.

7. A method for selecting a set of redundant connections for insertion in an integrated circuit (IC), the set of redundant wires capable of tolerating open faults in the IC, the method comprising:

generating, by a graphical construction unit, at least one graph, each graph corresponding to arrangements of pins and connections of a net in the IC, the graph representing a pin of the net as a vertex, and a connection between two pins as an edge;

identifying a first and a second pair of unconnected vertices in a graph for inserting a first and a second redundant edge, respectively, the first redundant edge and the second redundant edge forming a first connected loop and a second connected loop, respectively, each loop further including at least two edges of the graph;

identifying a different first pair of vertices if the second loop includes all of the at least two edges of the first loop, or a different second pair of vertices if the first loop includes all of the at least two edges of the second loop; and

selecting as the set of redundant connections, the connections corresponding to the first and second redundant edges of each graph;

calculating a tolerance ratio for the first redundant edge and the second redundant edge;

sorting the first and second redundant edge based on their tolerance ratio;

calculating a yield rate change of the first IC layout associated with inserting one of the first or second redundant edge with a highest tolerance ratio, and updating the first IC layout to include the redundant edge with the highest tolerance ratio if it is feasible and if the yield rate change is greater than zero; and

calculating the yield rate change of the first IC layout associated with inserting the first or second redundant edge having a second highest tolerance ratio, and updating the first IC layout to include the redundant edge with the second highest tolerance ratio if it is feasible and if the yield rate change is greater than zero.

8. The method of claim 7 , wherein generating a graph comprises:

calculating at least one projection point corresponding to a first vertex of a net in the graph, the at least one projection point being calculated as a shortest distance from the first vertex to at least one neighboring edge in the net; and

generating a projection graph by modifying the graph to include an additional vertex corresponding to the at least one projection point.

9. A system for manufacturing an integrated circuit (IC) capable of tolerating random manufacturing defects, the system comprising:

a graphical construction unit (GCU), the GCU configured for generating a graph corresponding to a net of an IC, the graph representing a pin of the net as a vertex, and a connection between two pins as an edge;

a redundant wire generation unit (RWGU) coupled to the GCU, the RWGU configured for identifying from a generated graph, a first and a second pair of unconnected vertices in the graph for inserting a first and a second redundant edge, respectively, the first redundant edge and the second redundant edge forming a first connected loop and a second connected loop, respectively, each loop further including at least two edges of the graph;

a router unit (RU) coupled to the RWGU, the RU configured for finding a feasible path associated with the first and the second redundant edge;

a yield computation unit (YCU) coupled to the RWGU, the YCU configured for calculating and sorting the first and second redundant edge based on their tolerance ratio, the YCU further configured for calculating a yield rate change associated with the first and the second redundant edge; and

a redundant wire insertion unit (RWIU) coupled to the YCU, the RWIU configured for implementing redundant connections in the IC based on the calculations of the YCU.

10. The system of claim 9 , wherein the GCU is further configured to:

calculate at least one projection point corresponding to each vertex of the graph as a shortest distance from the vertex to at least one neighboring edge; and

generate a projection graph by modifying the graph to include an additional vertex corresponding to the at least one projection point.

11. The system of claim 9 , wherein calculating the tolerance ratio comprises calculating the tolerance ratio of a redundant edge as

C

r



r



.

,

wherein |r| is a Manhattan length of the redundant edge, and C r is the total wire length of the connected and closed path formed by the redundant edge including the Manhattan length of the redundant edge.

12. The system of claim 9 , wherein calculating the yield rate change comprises calculating the yield rate change associated with a redundant edge as ΔY≅(C r −|r|)·o−ΔP′·s. wherein ΔY is the yield rate change associated with the redundant edge, |r| is a length of the feasible path corresponding to the redundant edge, C r is an increase in length of a feasible path corresponding to a connected path formed by the redundant edge, o is an open rate per unit wire length associated with the IC, ΔP′ is a change in parallel run length associated with inserting the redundant edge, and s is a short rate per unit parallel run length of wire in the IC.