SQL Server : is it possible to create non-clustered index for IS NULL condition

There is no problem creating indexes on nullable columns.
The only thing you can't do with nullable columns is create a primary key.

You can right click the green message and SSMS will give you the option to copy it. Then paste it into the query editor and you will see it contains the create index statement. All you need to do is give it it's name and run the statement.

Following the conversation in the comments to this answer, Ivan Sivak have figured out that this is a bug that SQL Server had in and was fixed in 2012 version. Here is the link to the bug report in Microsoft Connect.

By default, the query plan uses the best index in available in any scenario.

One reason could be that the index is fragmented. You could decrease the fragmentation by running the query

ALTER INDEX [myindexname] ON [dbo].[mytable] REBUILD;

You can read more on this here: https://learn.microsoft.com/en-us/sql/relational-databases/indexes/reorganize-and-rebuild-indexes

The query plan at times doesn't take the best available index for the query. For this, Microsoft has provided table hints.

1. ForceSeek
   This tells the optimizer to use an index seek operation in any case.

SELECT *
FROM   mytable WITH(FORCESEEK)
WHERE  mycolumn IS NULL 

Caution: while using any Force Plan hint (ForceSeek,ForceScan,etc.) or you might break the query

2. Index Hint
   This tells the optimizer to use a particular index for seek/scan

SELECT *
FROM   mytable WITH([myindexname])
WHERE  mycolumn IS NULL 

This query will now use myindexname for above query.

3. You can combine both these hints:

SELECT *
FROM   mytable WITH([myindexname], ForceSeek)
WHERE  mycolumn IS NULL 

This query will now use myindexname with a scan operation for above query.

More on table hints here: https://technet.microsoft.com/en-us/library/bb510478(v=sql.105).aspx