Is a "headphone amplifier" what I need? If so, any suggestions on which one?

Are you sure you've exhausted the options with your Scarlett's software? If your input gain is loud enough (but without clipping naturally) then you should be able to get enough gain at the headphone output, although I have no experience of Scarletts.

Having said that, I have an RME Babyface and I've heard people complain that the headphone outputs are too quiet. Can't understand it myself, because I find that I get to the level past which I would consider the level damaging to my hearing quite easily. I could never understand why anyone would want it louder. Pete Townsend ruined his hearing by listening to headphone mixes too loud in the studio.

Cool, mang. Honestly my brain is on fire with annoying bureaucratic bullpizzle at the moment so it was cathartic unleashing all that. I'm just glad you found it helpful.
 
Cheers. ;-)

Well to be honest even though the MixControl stuff is supposed to be "easy" to configure it always confused me and I found other ways to get things done anyway. It mostly seems more useful for if you need to send different mixes to different band members while they track simultaneously.
 
The only times I open it are to check input levels (because I find even if Sonar isn't clipping the MixControl may be) and to change my ASIO buffer settings.
 
I will say that, if you don't already, it is EXTREMELY helpful to have busses for ALL instruments. Even if you only have one bass track or one guitar track or whatever having those routed to a bus makes so many things so much easier. I never used busses before I got Sonar (didn't even really know what the heck they were) and good lord did it ever make things unnecessarily difficult.

Oh and Beepster, there's a huge difference between the entry level Scarletts and the Firewire models as far as RTL performance goes.    
 
The 2i2 -2i4 6i6 ( 18i4?)  have unacceptable round trip latency for real time monitoring. I have about 32 ms RTL.  But I don't use the zero latency setting either. I'm not sure what difference it makes. I'm using what ever the dang thing defaulted to when I installed it and rarely pay attention to it unless I'm fussing with the mike input headphone thing. 
 
Bottom line is, unless you toggle the input echo in Sonar and listen to what's passed though all those converters and stuff, your using zero latency monitoring because your headphones / monitors are patched directly to the inputs. If you turn on input echo in Sonar you will hear a delay. All interfaces work this way. 
 

Some great advice here.
 
I have one other suggestion.
 
 
SING LOUDER!!!!!!!

Kerch
...I have a set of headphones...

You haven't said which headphones you are using. I'm just wondering if maybe it's the headphones that are the problem.

Cakealex - one thing- 
 
"and you can a VST compressor into the chain (no need to fork out for a real compressor)." 
 
This is not true, You cannot compress before you hit the A/D with a VST compressor which is the why some of us work this way in the first place. It is one of those topics that can start a long discussion because it falls under the category of "technique". But there's a goodly number of engineers who will put a compressor on the pre amp to catch any overs. Once an over has hit your A/D your hooped. The argument will always go, well then just turn it down...but myself I have worked this way since the 4 track tape deck days and it works for me. A fast attack and a very low ratio will catch explosives and bad mike technique. And without actually compressing the vocal. 
 
People get confused about the Zero Latency monitoring and there was just a thread about it. Simply put, all interfaces have a way to listen to your input either in the cans or up on the monitors. You then blend in the output of the DAW  and go at it. All DAWS adjust the playback so that overdubs are in sync. That's why your DAW needs to calculate the latency. 
Everyone should take the time to test it's accuracy. 
 
Ya it's to bad the OP ordered the Headphone Amp before waiting for a few more replies. Not a bad thing to have as they are handy for recording bands and groups. The Scarlett has 2 (loud) Headphone amps built in but you would want the outboard unit for using the rear outputs for a cue mix. 

sock monkey
Cakealex - one thing- 
 
"and you can a VST compressor into the chain (no need to fork out for a real compressor)." 
 
This is not true, You cannot compress before you hit the A/D with a VST compressor which is the why some of us work this way in the first place.

That is true my bad - I was typing when fast asleep.~ The rest of the post was accurate though.
This reminds me I need to get a cloudlifter for my SM7B to increase it's gain. Works well with the focusrite but I definitely would like more gain ideally.

Thanks for all the replies, guys. I did purchase the ART mymonitor, received it yesterday, and i'm really liking it so far with the minimal time I've been able to spend on it. Luckily I can easily return it if I'm able to find some way to finally get what I wanted within the Scarlett itself, so I'll be using some of these suggestions today and fooling with it one last time. 
 
As to some of the specifics asked, I'm using an AKG C214 mic, along with AKG headphones. Can't remember the model of those phones; they're not top of the line, not terrible either. Decent mid-range pair. 
 
I think my issues comes down to the gain I've been setting on my mic. When I set it initially, I would carefully watch the meters as I sang some test passages, and everything stayed in the green, no yellow/red. But when I would play back. louder passages would have a sort of clipped sound, almost as if someone had applied a way too-heavy brick wall compression on it. I had a pad that came with the mic switched on while singing. 
 
Anyway, I found I could remedy this by lowering the gain volume to the point where the green level indicator on the Scarlett barely comes on while singing these sections. The result is that the vocal doesn't have that "overly compressed" or distorted sound, but the result of THAT has been that the levels in the mix have been too low for me to really hear well, even with everything in Sonar's mix control cranked to full volume for the vocal and lowered as much as possible for the music mix. I tried to remedy that by applying input echo, getting the latency low and adding some compressor to the back end in Sonar while singing, and that's helped, but has also brought some other issues into the mix (pardon the pun), i.e., not totally comfortable now with the processed sound of the vocal coming through. I know, nitpickery. 
 
I think a lot of it comes from how I sing certain parts - very loud, very hard, and my songs tend to usually have some pretty big changes in dynamics. I've already done separate tracks for a "softer" verse and a "harder" chorus, so I'm not singing both of these parts on the same track during the same take. 
 
So yeah, maybe there's a technique thing involved that I need to work on, but I do know enough as far as when to back off the mic, get up on the mic, etc. In any case, thank you for all the great replies. I will have some time this afternoon to fool with this some more, but hopefully the above explains my situation a little more clearly. 

That is definitely the type of thing that a) would require some kind of input limiting/compression (which is destructive and I personally think should be done AFTER tracking unless you've got some REALLY nice output gear and know how to use it), b) do multiple takes as you said, one with your input level set high enough to get a healthy signal for the quiet parts and a second one at a lower level so the loud parts don't clip your input channel (notice I said input channel... NOT the track in Sonar... I will explain further later) or c) go old school and have someone in the room riding the input fader (they would have to know the parts so they know when to do it and how far they have to go down in volume which is a simple calculation but who wants someone else in the room... I know I don't, especially when I'm signing... lol). Of course you could do that last one yourself but that might distract from your performance.
 
Now as far as overloading the channel input versus Sonar... the ONLY place you really need to worry about clipping is at the interface (if you aren't using any other outboard stuff like a preamp). Even if the clip lights in the track are going off it doesn't matter if the clip lights are staying out of the red on the interface.
 
The clip lights in Sonar aren't showing actual clipping that will be recorded (until when you mix down... then you may get digital distortion which is definitely unpleasant). What Sonar WILL show when you have recorded a clipped input is a white vertical line printed on the waveform. If you aren't getting any of those and you aren't HEARING any distortion/clipping then it does not matter what the track meters say.
 
From there you can simply automate if need be in the mix to even the levels out.
 
So do not look at the track faders in Sonar. Look at your interface LED's (which unfortunately can be a little inaccurate and they don't hold peaks) and right on the waveform for those thin vertical white lines. Most importantly LISTEN for clipping because that's all that really matters anyway.
 
I say this because you are saying you are getting yellow and red lights in Sonar. I get that quite often when tracking because I want a good strong signal and I know that as long as I'm good at the interface and I'm not seeing those white lines/hearing anything.
 
I used to be afraid of the track meter lights in Sonar and adjusted my input to avoid them. The result was really weak signals to work with which sucks. Things are much better now.
 
You may know all that but thought I would mention it.
 
Also I'm still learning so if I made any mistakes maybe some of the more experienced guys can correct me.
 
Cheers.

Kerch
Thanks for all the replies, guys. I did purchase the ART mymonitor, received it yesterday, and i'm really liking it so far with the minimal time I've been able to spend on it. Luckily I can easily return it if I'm able to find some way to finally get what I wanted within the Scarlett itself, so I'll be using some of these suggestions today and fooling with it one last time. 
 
As to some of the specifics asked, I'm using an AKG C214 mic, along with AKG headphones. Can't remember the model of those phones; they're not top of the line, not terrible either. Decent mid-range pair. 
 
I think my issues comes down to the gain I've been setting on my mic. When I set i hit initially, I would carefully watch the meters as I sang some test passages, and everything stayed in the green, no yellow/red. But when I would play back. louder passages would have a sort of clipped sound, almost as if someone had applied a way too-heavy brick wall compression on it. I had a pad that came with the mic switched on while singing. 
 
Anyway, I found I could remedy this by lowering the gain volume to the point where the green level indicator on the Scarlett barely comes on while singing these sections. The result is that the vocal doesn't have that "overly compressed" or distorted sound, but the result of THAT has been that the levels in the mix have been too low for me to really hear well, even with everything in Sonar's mix control cranked to full volume for the vocal and lowered as much as possible for the music mix. I tried to remedy that by applying input echo, getting the latency low and adding some compressor to the back end in Sonar while singing, and that's helped, but has also brought some other issues into the mix (pardon the pun), i.e., not totally comfortable now with the processed sound of the vocal coming through. I know, nitpickery. 
 
I think a lot of it comes from how I sing certain parts - very loud, very hard, and my songs tend to usually have some pretty big changes in dynamics. I've already done separate tracks for a "softer" verse and a "harder" chorus, so I'm not singing both of these parts on the same track during the same take. 
 
So yeah, maybe there's a technique thing involved that I need to work on, but I do know enough as far as when to back off the mic, get up on the mic, etc. In any case, thank you for all the great replies. I will have some time this afternoon to fool with this some more, but hopefully the above explains my situation a little more clearly. 

There is:

Mic gain on your focusrite. Make sure the levels are right for recording. Check the levels for recording within Sonar! Ignore the levels you hear!

Then....

Gain/trim knob for track in Sonar. Are you making full use of this on playback?

Volume (fader)/for track in Sonar. Generally you keep it around the default setting when recording.

And you have the mic volume for monitoring in mix control IF you are using the zero latency feature.

The levels you record and the levels you monitor whilst recording should be viewed upon separately.

Consider putting a VST compressor into a track for playback purposes along with your reverb effect.

" as I sang some test passages, and everything stayed in the green, no yellow/red. But when I would play back. louder passages would have a sort of clipped sound, almost as if someone had applied a way too-heavy brick wall compression on it."
 
Here seems to be telling the story... If the Scarlett and Sonar do not show clipping yet you GET clipping, this points at something wrong with the mike.