As part of an ongoing program to better understand the early stages of massive star cluster evolution and the physical conditions for their formation , we have obtained J , K ^ { \prime } , and N ( 10.8 \mu m ) images of the nuclear region of the starburst galaxy He 2-10 .
The N -band images were obtained with the Gemini North telescope .
In only ten minutes of on-source integration time with Gemini we were able to detect three of the five enshrouded clusters , or “ ultradense H II regions ” ( UDH II s ) recently discovered in radio maps .
None of these sources appears in either the optical HST images or the near-infrared ( J , H , and K ^ { \prime } ) images .
These sources comprise about 60 % of the total N -band flux from He 2-10 and , we suspect , a similar fraction of the total far infrared flux measured by IRAS .

The inferred spectra of the UDH II s are strikingly similar to those of Galactic ultracompact H ii regions .
We have modeled the radio+IR spectrum of these UDH II s under the assumption that they are “ scaled-up ” Galactic ultracompact H ii regions .
From this model , the bolometric luminosity of the brightest cluster alone is estimated to be \approx 2 \times 10 ^ { 9 } L _ { \odot } .
The total mass of the dust and gas in this UDH II is M _ { shell } \approx 10 ^ { 7 } ~ { } M _ { \odot } .
We have also used the observed spectra to place constraints on the masses and ages of the stellar clusters enshrouded within the UDH II s. For the brightest UDH II , we find that the stellar mass must be M _ { Cluster } \gtrsim 2.5 \times 10 ^ { 6 } M _ { \odot } and the age must be \lesssim 4.8 \times 10 ^ { 6 } yr , with the most probable age \lesssim 3.6 \times 10 ^ { 6 } yr .
If we assume that the region is pressure confined and enforce the requirement that the star formation efficiency must be less than \sim 90 % , we find that the age of this stellar cluster must lie within a very narrow range , 4 \times 10 ^ { 5 } < \tau < 5 \times 10 ^ { 6 } yr. All of the clusters within the UDH II s in He 2-10 are estimated to have ages less than about 5 \times 10 ^ { 6 } yr and masses greater than about 5 \times 10 ^ { 5 } ~ { } { M } _ { \odot } .

We find that the logarithmic ratio of the radio to far-infrared flux densities , q , for the UDH II s in He 2-10 is \sim 4 ; q \approx 2.6 for both He 2-10 as a whole and NGC 5253 , another nearby starburst known to host UDH II s. These values of q are significantly larger than the average q = 2.35 found for normal galaxies , but comparable to the values of q found for ultraluminous infrared galaxies .
We suggest that large q values for starburst galaxies may indicate that a significant fraction of the far-infrared flux may arise from thermal dust emission from UDH II s .

Finally , the possibility that all of the far-infrared flux from He 2-10 and other starburst galaxies may be produced by regions completely obscured at wavelengths as long as K ^ { \prime } suggests that the well-known correlation between ultraviolet continuum slope and infrared-to-ultraviolet flux ratio in starbursts can not be due entirely to reprocessing of ultraviolet radiation by dust in a foreground screen geometry .
In fact , the dust which reddens the ultraviolet continuum slope must be largely decoupled from the dust which produces the large infrared fluxes in some starbursts .