Radial profiles become more complex when changes in stellar colours and ages are considered. Bakos et al (2008) noted that Type II light profiles tend to correspond to U-shaped B − V colour profiles, which means that the inner part of the disk gets bluer with radius at first, and then the outer part of the disk gets red again. This colour change presumably corresponds to a change in the mass-to-light ratio, with large ratios in the outer parts. Then the down-bending Type II in a light profile tends to straighten out and become Type I in a mass profile. That is, the outer red trend gives an increasing mass-to-light ratio, causing an increasing conversion factor from surface brightness to mass surface density. The red outer parts could be from old stars that scattered there from the inner regions (Roškar et al 2008), as distinct from the common model of inside-out growth for spiral galaxies. A larger survey recently confirmed this result. Zheng et al (2015) included 700 galaxies using deep images from the Pan-STARRS survey. The average g-band (peak at 5150 Å) light profile was the down-bending Type II for low-mass galaxies (< 1010 M⊙) and slightly less bent for high mass galaxies (< 1010.5 M⊙), as usual, and the average g−i colour profiles (i band peaks at 7490Å) were U-shaped to various degrees, so the average mass profile became Type I for all galaxy masses. Muñoz-Mateos et al (2015) made average radial profiles separated into eight mass bins for ∼ 2400 galaxies using 3.6 µm emission from the Spitzer Survey of Stellar Structure in Galaxies. Such long wavelength emission is a nearly direct probe of galaxy mass, although there is some PAH emission from dust in it too. All masses showed Type II profiles on average, with a straighter trend like Type I from a central bulge in the more massive galaxies. This suggests that the mass profile for most galaxies is not exactly Type I, but still tapers off more steeply in the outer parts, beyond 1 kpc for low mass galaxies (< 109 M⊙) and beyond 10 kpc for high mass galaxies (> 1010.5 M⊙).
The most telling observations are of stellar age gradients because colour gradients can be from a mixture of age gradients and metallicity gradients. Roediger et al (2012) determined radial age profiles from photometry and stellar population models of 64 Virgo cluster disk galaxies. They found U-shaped age profiles in 15% of Type I's, and also in 36% of both Types II and III. In one-third of all exponential types, the age increased steadily with radius. Yoachim et al (2012) found about the same mixture of age profiles, measuring ages from spectra in 12 galaxies. Dale et al (2016) determined star formation histories for 15 nearby galaxies with masses in the range 108 M⊙ to 1011 M⊙ using ultraviolet and infrared data; they also found U-shaped age profiles. These results imply that outer disks generally have old stars, although most also still have some star formation.
A recent integral field unit survey of 44 nearby spiral galaxies (CALIFA) by Ruiz-Lara et al (2016) also found U-shaped age profiles in Types I and II when the stars were weighted by brightness, as would be the case from integrated spectra or photometry. This is in agreement with the previous surveys mentioned above. The galaxies were observed beyond their break radii or for at least three scale lengths. In contrast, Ruiz-Lara et al (2016) found constant age profiles when the stars were weighted by mass. They suggested that the entire disk formed early with star formation stopping in the inner parts first, and then quenching from inside-out. This is unlike cosmological simulations that have the outer disk form more slowly than the inner disk, and also unlike models where the outer disk stars migrate there from the inner disk.
Watkins et al (2016) viewed three nearby spirals with very deep images, covering a range of about 10 magnitudes in surface brightness for B band. They found smooth and red stellar distributions with no spiral arms in the far-outer disks. For the typical colour of B − V ∼ 0.8 mag in the outer parts, and from a lack of FUV light, they concluded that the star formation rate (SFR) had to be less than 3−5 × 10−5 M⊙ pc−2 Myr−1. This seemed to be too low for continuous star formation and disk building, suggesting some radial migration. However, the lack of spiral arms makes the usually invoked churning mechanism (Sellwood and Binney 2002, Roškar et al 2008, Berrier and Sellwood 2015) inoperable. Churning is a process of stellar migration back and forth around corotation. Perhaps stellar scattering off local gas irregularities makes the outer exponential structure (Elmegreen and Struck 2013).