Fermi Non-detections of Four X-Ray Jet Sources and Implications for the IC/CMB Mechanism

One problematic aspect of the IC/CMB model is that it is energetically costly, with kinetic jet powers at or sometimes exceeding the Eddington luminosity for these sources (Dermer & Atoyan 2004). This is primarily because it requires extending the electron energy distribution to very low energies to provide electrons of the right energy to produce the observed X-ray emission from upscattered infrared CMB photons. This low-energy extension (i.e., a low ${\gamma }_{\min }$ value) of the electron energy distribution greatly increases the total energy content of the jet by requiring a much larger total number of electrons. While such an extension is possible, there is no direct evidence for these low-energy electrons because their synchrotron radiation would be at frequencies much less than our GHz-frequency radio observations.

The radiative losses of the electrons responsible for the X-ray emission are very weak in the IC/CMB model, with radiative lifetimes in excess of 10⁶ years (Harris & Krawczynski 2006). In the simplest case, we would thus expect continuous X-ray emission along the entire jet length rather than the observed X-ray knots noted by Atoyan & Dermer (2004) and Stawarz et al. (2004). This problem can be avoided if the X-ray emitting plasma is confined to moving blobs rather than a continuous jet flow (Tavecchio et al. 2003). This solution would require these X-ray emitting blobs to be propagating with a high Γ, which may be observable with proper motion studies for nearby sources. However, in the case of 3C 273, Meyer et al. (2016) found that the knots are stationary, with a limit of apparent speed ${\beta }_{\mathrm{app}}\ \lt \ 1c$, corresponding to a limit of ${\rm{\Gamma }}\ \lt \ 2.9$, which is incompatible with the IC/CMB interpretation for the X-rays.

A feature of some of these MSC jets is offsets in peak brightness between radio, optical, and X-ray wavelengths when observed with similar angular resolutions (see the case of 3C 111 in Clautice et al. (2016) for a clear example of this). There is no reasonable explanation for these offsets in the IC/CMB model because they are inconsistent with the idea that the X-ray, radio, and optical emission is produced by the same population of relativistic electrons, where we would expect co-spatial emission regions.

In addition to the above difficulties, there have been several observations of MSC jets that directly challenge the IC/CMB model. Hubble Space Telescope (HST) polarization measurements of the rising UV-component in the jet of PKS 1136–135, clearly part of the second spectral component, show fractional polarization measures in excess of 30% for several knots (Cara et al. 2013). Because IC/CMB emission is expected to have very low polarization (see Uchiyama 2008; McNamara et al. 2009), it is extremely unlikely that the second component is produced by the IC/CMB mechanism in this source. Another recent study by Hardcastle et al. (2016) of the nearby FR II radio galaxy Pictor A shows IC/CMB to be incompatible with the observed jet to counter-jet flux ratio. They found a ratio orders of magnitude less than what would be expected if IC/CMB were the source of the X-rays (in which case, ${\rm{\Gamma }}\geqslant 5$ and a jet angle less than a few degrees is required). While the conclusions drawn from these observations of PKS 1136–135, 3C 273, and Pictor A do not necessarily apply to all MSC jets, they already necessitate a second emission mechanism, other than IC/CMB, to explain the X-ray emission in these cases.

There are several alternative models that might explain the MSC jets. In particular, a second population of synchrotron-emitting electrons (Atoyan & Dermer 2004; Harris et al. 2004; Kataoka & Stawarz 2005; Hardcastle 2006; Jester et al. 2006; Uchiyama et al. 2006) and various hadronic emission models (Aharonian 2002; Kusunose & Takahara 2017; Petropoulou et al. 2017). Until recently, it has been difficult to reject the IC/CMB model in favor of any alternative, because in most cases any model can be tuned to reproduce the somewhat sparsely sampled spectral energy distribution (SED). However, the question of which emission process is at work in these X-ray jets is important because these models imply vastly different physical properties.

In the case of the second synchrotron model, it will be important to understand why there are multiple electron energy distributions and how this is connected to the particle acceleration taking place in these jets. An immediate consequence of a higher-energy electron distribution is subsequent TeV gamma-rays produced when these electrons IC scatter the CMB (Georganopoulos et al. 2006; Meyer et al. 2015). These TeV gamma-rays may be detectable for low-redshift sources with the upcoming Cherenkov Telescope Array (CTA). Alternatively, if the source of the X-rays is due to hadronic emission processes, then this has important implications for the particle make-up and energy content of these jets.

As noted by Georganopoulos et al. (2006), the IC/CMB model predicts a high level of IC/CMB gamma-ray emission from the jet, which should be detectable with the Fermi/LAT. In previous work, we developed a technique to search for the steady gamma-ray IC/CMB emission component at times when the variable core was in a low state, necessary because the resolution of Fermi is insufficient to separately resolve the core and jet. This technique was applied in 3C 273 (Meyer & Georganopoulos 2014) and PKS 0637–752 (Meyer et al. 2015, 2017), where in both cases the upper limits to the gamma-ray flux during times when the core was quiescent were well below the required level from IC/CMB. This paper follows these works for four more MSC jets where the X-ray emission has been previously modeled as due to IC/CMB. However, in this paper we have targeted sources which are not detected in the Fermi/LAT 3FGL four-year point source catalog, which strongly suggests that the gamma-ray emission expected under the IC/CMB model may be absent, without any complication of gamma-ray emission from the core. Additional MSC jets are currently being analyzed in order to establish which sources may be consistent with the IC/CMB model based on gamma-ray observations, but the presence of newly detected point sources in the region of interest (ROI) makes the analysis more complicated, and we leave the discussion of these sources to a future paper.

In Section 2, we present the new and archival radio and optical data analyzed for our four targets, as well as the method of deriving Fermi upper limits. In Section 3, we present the resulting limits, which clearly violate the expectations of the IC/CMB model. In Section 4, we discuss the properties of the small but growing sample of MSC jets in which IC/CMB has been ruled out, including several lines of evidence that suggest they are not extremely well-aligned, as required under the IC/CMB model. Finally, in Section 5 we summarize our findings.

In Table 1, we summarize the source properties of the four targets presented in this paper and the two sources in which the IC/CMB model for the X-rays has already been ruled out, PKS 0637–752 and 3C 273. In columns 2–4, we report the source redshift and corresponding angular size scale in kpc per arcsecond, and the black hole mass as M_BH, taken from the literature where available (as noted in the table). Using data from NED, we decomposed the radio spectrum of these sources into the extended (isotropic) "lobe" component and beamed "core" component as in Meyer et al. (2011) and shown in Figure 1. In this figure, we plot the lobe emission with cyan squares and the core emission with blue circles (data filtered out from the fitting is shown as open circles). The lobe and core emission were modeled with power-law and log-parabolic spectra, respectively, shown as the dotted lines, with the combined fit shown as a red line. In column 5, we give the kinetic jet power L_kin, where we have scaled from the 300 MHz radio luminosity of the lobes using the scaling of Cavagnolo et al. (2010). In column 6, we give the radio core dominance R_CE, where this is the logarithmic ratio of the core to extended (i.e., lobe) spectrum at 1.4 GHz. In column 7, we give the radio crossing frequency ${\nu }_{\mathrm{cross}}$, where this is the frequency that the core spectrum crosses with the lobe spectrum. Also following the methods described in Meyer et al. (2011), we made a simple phenomenological fit to the synchrotron spectrum in order to determine an approximate peak luminosity and frequency (L_peak and ${\nu }_{\mathrm{peak}}$ respectively), which are given in columns 8 and 9 with the corresponding SED model fits shown in Figure 2. For the complete SEDs, we plot the lobe spectrum as cyan squares, Fermi/LAT data in black, the rest of the data with blue circles (data filtered out from the fitting is shown as gray circles), and the model fits as solid red lines. Finally, we give the projected size of the source and equipartition magnetic field strength for the brightest X-ray knot, B_eq, in columns 10 and 11. It should be noted that all of these sources have high kinetic jet powers (see Figure 4 from Meyer et al. 2011) and black hole masses (see Figure 1 from Chiaberge & Marconi 2011) relative to other radio-loud AGN in general.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 1.  Source Properties

(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)
Source	z	kpc/''	log ${M}_{\mathrm{BH}}/{M}_{\odot }$	log Lkin	RCE	log ${\nu }_{\mathrm{cross}}$	log $\nu {L}_{\nu ,\mathrm{peak}}$	log ${\nu }_{\mathrm{peak}}$	Size	Beq
Name				(erg s−1)		(Hz)	(erg s−1)	(Hz)	(kpc)	(10−4 G)
0637–752	0.650	6.75	9.41 (a)	45.9	0.19	9.01	46.3	13.4	164	1.15
1136–135	0.560	6.26	8.45 (b)	45.9	−0.72	10.7	45.4	14.1	92.9	0.470
1229–021	1.05	7.91	8.70 (b)	46.0	0.14	9.00	46.1	14.1	149	3.20
1354+195	0.720	7.03	9.44 (a)	45.8	−0.50	9.50	46.0	13.6	305	2.07
2209+080	0.480	5.80	⋯	45.5	−1.5	10.4	45.6	13.4	64.4	1.00
3C 273	0.160	2.65	8.91 (c)	45.5	0.37	8.82	45.9	14.4	63.6	1.00

Note. Size of the source is the projected size determined from hot spot to counter hot spot. (a), (b), and (c) black hole masses taken from Chen et al. (2015), Liu et al. (2006), and Kim et al. (2015), respectively.

Download table as:  ASCIITypeset image

In this paper, we aim to test whether the high level of steady gamma-ray emission predicted under the IC/CMB model for the X-rays is seen. It is therefore essential that the precise level of the expected gamma-ray emission is known. As discussed in Georganopoulos et al. (2006), the X-ray to gamma-ray IC component has the same spectral shape as the radio-optical synchrotron spectrum, and the requirement to match the observed X-rays thus fixes the rest of the spectrum which peaks in the gamma-rays. It is thus critical that we compile the best possible multi-wavelength SEDs for the knots, from radio to X-rays. We describe here the archival and new data used to compile these SEDs for our four sources.

As shown in Table 2, we used the deepest available archival VLA imaging of sufficient quality in bands L, C, X, and U (A or B configuration) to obtain a high-resolution image with the knots distinctly resolved. All sources were analyzed using CASA version 4.7.0. In all cases, either 3C 286 or 3C 48 was used as the flux density calibrator, and the source itself was used for gain calibration. We applied several rounds of self-calibration before generating the final image with clean. In Table 2, we list the source, band and configuration, central frequency of the observation, project code, date of observation, time on source (TOS) in minutes, rms of the final self-calibrated image in μJy, and size of the restoring beam in arcseconds.

We searched for archival VLBI observations of our targets that can be used to measure the pc-scale jet proper motions. Three of the targets have no suitable archival data or were not observed with VLBI over multiple epochs. Only one target, PKS 1354+195, was observed four times (in 1997, 1999, 2002, and 2003) as part of the Monitoring Of Jets in AGNs with Very Long Baseline Array (VLBA) Experiments (MOJAVE) monitoring project (Lister et al. 2009, 2016), although no previous proper motion measurements were reported. We downloaded the fully-reduced fits images from the MOJAVE website and used the publicly available Wavelet Image Segmentation and Evaluation (WISE) code (Mertens & Lobanov 2015, 2016) to analyze the images for proper motions.

The WISE code comprises three main components. First, detection of structural information is performed using the segmented wavelet decomposition method. This algorithm provides a structural representation of astronomical images with good sensitivity for identifying compact and marginally resolved features and delivers a set of two-dimensional significant structural patterns (SSP), which are identified at each scale of the wavelet decomposition. Tracking of these SSP detected in multiple-epoch images is performed with a multi-scale cross-correlation algorithm. It combines structural information on different scales of the wavelet decomposition and provides a robust and reliable cross-identification of related SSP.

The images of PKS 1354+195 were taken with the VLBA at 2 cm (15 GHz) on 1997 August 18, 1999 July 19, 2002 August 12, and 2003 January 06. The images were centered on the core position and required no additional adjustment before analysis with WISE. The final reported proper motion values were obtained using a scale factor of six, though results were similar across a range of reasonable scale factors for the size of the knots in the jet. We only report those features detected in at least three of four epochs.

In Table 3, we list the ALMA data used in our analysis, where we give the observing band, cycle of the observations, effective frequency of the flux measurement, TOS, rms of the final image, and size of the restoring beam. ALMA observations were made of the first three targets as part of Cycle 3 and 4 programs 2015.1.00932.S and 2016.1.01481.S. Both programs included observations at band 3 and 6; however, only about 40% of the C-rated cycle 3 program was completed, and not all (A-rated) cycle 4 observations have been released. Forthcoming observations may improve on those presented in this paper. PKS 1136–135 was observed in band 3 by ALMA as a calibrator for a project unrelated to jets (program 2016.1.01250.S, PI: C. Peroux), and this data was kindly made available to us by the PI. For all sources, the data was first reduced using the provided scriptForPI.py script, which produces a calibrated measurement set appropriate for imaging with clean. The source scans were split off and several rounds of phase-only (non-cumulative) self-calibration were applied before a final (cumulative) round of amplitude and phase self-calibration. Clean was used in mfs mode, with nterms = 1 and Briggs weighting with robust = 0.5. The final rms measurements correspond to the primary-beam-corrected image, though we use the uncorrected images in the figures. In the case of 2209+080, we combined and imaged the somewhat lower-resolution cycle 4 data with the cycle 3 data, after core subtraction.

Archival HST flux densities for the individual knots of 1354+195 were taken from Sambruna et al. (2004), from Cara et al. (2013) for 1136-136, and from Tavecchio et al. (2007) for 1229–021.

PKS 2209+080 was observed as part of our HST program GO-13676 (PI: Meyer) and we present the results here for the first time. This source was observed for one orbit with the Wide Field Camera 3 (WFC3) in the near-infrared (IR) channel with filter F160W (λ = 1.6 μm), and for two orbits with the Advanced Camera for Surveys Wide Field Camera (ACS/WFC) with filter F606W (λ = 6000 Å). In both cases, a dithering pattern was used to better sample the PSF, and the raw images were stacked using a routine similar to AstroDrizzle, but better optimized for Astrometry. In particular, several dozen point sources in the optical image were used as a reference frame on which all images were aligned (final error on the registration is less than 1 mas). This allowed us to precisely align the optical and IR images, which allows for better identification of common features in both. The final pixel scale in both images is 25 mas.

We did not attempt a galaxy subtraction, as it has a very irregular profile and models produced by iraf tasks ellipse and bmodel left large residual errors. We measured flux densities using contours around each knot, estimating the background by moving the same contour to 6–8 positions at the same radius from the galaxy center. The mean background value was subtracted from the initial flux density measurement, and errors are taken to be $\sqrt{2}$ times the standard deviation of the background flux densities. The flux densities were measured using the viewer utility of the CASA package.

By design, each of the four targets was not a member of the 3FGL 4 year point source catalog for the Fermi/LAT. We first checked if any of the sources would be detected using data for the entire time range since the launch of Fermi in 2008 (over eight years). Fermi/LAT event and spacecraft data were extracted using a 10^◦ ROI, an energy cut of 100 MeV–100 GeV, a zenith angle cut of 90^◦, and the recommended event class and type for point source analysis for all sources except PKS 1136–135, where a 7^◦ ROI was used. The time cuts included all available Fermi data at the time of analysis with corresponding mission elapse time (MET) ranges of 239557417 to 510106834 for PKS 1136–135, 239557417 to 501427132 for PKS 1229–021, 239557417 to 239557417 for PKS 1354+195, and 239557417 to 503840003 for PKS 2209+080. Following the standard methodology for Fermi/LAT binned likelihood analysis, a binned counts map was made with 30 logarithmically spaced energy bins and 0.2 degree spatial bins in all cases. Initial spatial and spectral model files for each source were constructed with sources up to 10^◦ outside the ROI, using the publicly available make3FGLxml.py script. This populates the model file with point and extended sources from the Fermi/LAT 3FGL catalog and an extended source catalog, respectively. Additionally, the current galactic diffuse emission model, gll_iem_v06.fits, and recommended isotropic diffuse emission model for point source analysis, iso_P8R2_SOURCE_V6_v06.txt, were used for the analyses. The livetime cubes were computed using 0.025 steps in $\cos (\theta )$ (where θ is the inclination with respect to the LAT's z-axis) and 1^◦ spatial binning. All-sky exposure maps were then computed using the same energy binning as the counts maps. After obtaining converged fits following the maximum likelihood optimizations with the initial model, Test Statistic (TS) residuals maps with 1^◦ spatial binning were created to find new point sources not accounted for in the 3FGL catalog, following the procedure outlined in Meyer et al. (2015). The TS is defined as twice the logarithmic ratio of the maximum likelihood calculated with an additional point source at the specified location to the maximum likelihood without an additional source, and can be taken as roughly the significance squared. In Figure 3, we show the TS residual maps created before and after adding the new sources not accounted for in the 3FGL catalog. The new source positions are shown with green circles, and the sources we obtained upper limits for this paper are shown with white circles.

Zoom In Zoom Out Reset image size

In locations where TS residual values were greater than 10, we added point sources with power-law spectra to the model files and optimized their positions and spectral parameters. After obtaining converged maximum likelihood fits for the updated model file, our target source of interest was added to the model file with a fixed photon index (1.93 for PKS 1136–135, 2.39 for PKS 1229–021, 1.94 for PKS 1354+195, and 1.65 for PKS 2209+080). These photon indices were derived from the implied gamma-ray index of the IC/CMB model. After this, the upper limits in the five Fermi energy bands of 100 MeV–300 MeV, 300 MeV–1 GeV, 1 GeV–3 GeV, 3 GeV–10 GeV, and 10 GeV–100 GeV were computed by running the analysis tools separately on each data set with the appropriate energy range data cuts.

PKS 1136–135 is a powerful FR II source previously observed by Sambruna et al. (2006) with Chandra and HST, where they modeled the X-ray emission with the IC/CMB model. Our multi-wavelength images are shown in Figure 5, where we see a straight and knotty radio jet, with knots brighter toward the hot spot in the radio. Knots B, C, D, and E are not separately resolved by ALMA, so we do not include ALMA flux densities for the jet SED. In Figure 4, we show the SED of knots α, A, and B, where our gamma-ray upper limits are well below the extrapolated IC/CMB emission in all three cases. Our limits were not deep enough to rule out IC/CMB for the last three knots, which had low IC/CMB predicted gamma-ray flux densities. In knots α and A, there is an upturn in the SED at 815 nm that lies above a single power-law fit for the synchrotron spectrum. Fitting these UV-upturns with an IC/CMB model is problematic because this would require very high levels of X-ray emission not observed by Chandra. Therefore, these UV-upturns imply a second higher-energy electron distribution also responsible for the X-ray emission, similar to that seen in 3C 273. Cara et al. (2013) showed, with HST polarimetry, that all of the jet knots had fractional polarization measures in excess of 30% (except knots α and B, where they found $2\sigma$ upper limits of 15% and 14% respectively). For the cases where the UV emission is clearly part of the second spectral component, these high degrees of polarization provide strong evidence against the IC/CMB model. However, our gamma-ray upper limits provide another independent line of evidence ruling out the IC/CMB model in knot A, while our upper limits show that the IC/CMB model is ruled out for knots α and B, which did not have high degrees of polarization. Fitting the UV data with an IC/CMB model for knots α and A would also predict much higher X-ray flux densities than observed, though the UV spectrum is also much harder than would be consistent with an IC/CMB model fit.

PKS 1229–021 was previously observed and modeled as an IC/CMB X-ray jet by Tavecchio et al. (2007), where we use their reported HST and Chandra densities in our SED (Figure 4). The radio images shown in Figure 6 show four well-defined cannonball-like knots downstream of the core. Interestingly, similar periodic knot structures are seen in all six of the jets discussed in this paper. As shown in Figure 6, knots B, C, and D are not separately resolved by ALMA and Chandra. Thus, we combined the data for these knots, labeled as knot BCD. Past this combined feature, we see the jet is considerably bent, implying a change of jet direction. Our ALMA images (which are core-subtracted) are focused on the jet in order to emphasize the knot structure. We found the western hot spot was only detected by ALMA in band 3, but not in band 6. The HST data presented in Tavecchio et al. (2007) shows that only the combined knot BCD was detected in the HST imaging, while knot A is not detected. With only one optical data point, the synchrotron SEDs for both of these features are not as well-constrained as for our other sources. Therefore, we believe that this source would be a good candidate for follow-up HST observations in order to further constrain the SED.

This source is shown in Figure 7, where the radio observations show a straight jet with detailed hot spot structure for the northern hot spot. The VLA and ALMA observations again show a well-defined cannonball-like knot structure for knots A and B. This jet was previously observed and modeled as an IC/CMB X-ray jet by Sambruna et al. (2002, 2004), so we use their reported HST and Chandra flux densities from the 2004 publication. Initially, we focused on knot A, which was reported as brightest in the X-rays. This knot is approximately 2'' from the core and initially reported in Sambruna et al. (2002) to have a flux density of 8.2 nJy (corresponding to about 135 counts) and later revised to 16.1 nJy in Sambruna et al. (2004). However, deeper Chandra observations of PKS 1354+195 suggest that knot A is in fact not detectable separate from the bright quasar core, with an upper limit of <0.39 nJy, after taking into account careful modeling of the PSF from the core (Harris et al. 2017). We plot the 8.2 nJy Chandra flux density in Figure 4 as a red open circle. While this flux density implies that the IC/CMB model is ruled out at a very high level of significance (see red dashed line), the much lower revised flux density limit (blue triangle and dashed line) does not. In light of this revision, we also examined the newer Chandra observations and can confirm that we do not find a significant excess of counts at the position of knot A in any of the later observations, suggesting that the apparent excess in the early observation (Chandra Observation ID #2104) was a statistical fluctuation. The jet of PKS 1354+195 is quite straight up to just before the hotspot, suggesting no major bends in or out of our line of sight. If knot A has the same radio to X-ray ratio as knot B, we would expect an X-ray flux density of about 2.4 nJy. However, the very bright core makes it difficult to conclude much from knot A. Turning to the remaining knots in the jet, and assuming some level of X-ray flux density from knot A, we produce the lower limit X-ray flux density shown as a green square in the SED figure with a value of 2.4 nJy. Note this is approximately the same X-ray flux density calculated for knot A, assuming the same radio to X-ray ratio as knot B. Using this and the spectral shape determined by knots A and B together from radio-optical, we see that the IC/CMB model is still ruled out, though the case is not as strong as in the other sources presented in this paper.

Note that, in Table 8, the predicted IC/CMB level, the 95% upper limits, and the corresponding confidence and limits on δ all correspond to the more reliable estimates derived from the knots excluding knot A.

3.3.1. VLBI Proper Motions

PKS 1354+195 was the only source for which archival VLBA data were available for a proper motion analysis. The results of the WISE analysis reveal three moving features in the jet. We show the pc-scale jet in the 2002 epoch and the results of the WISE feature tracking in Figure 9. Ordered by distance from the core, the moving features show angular motions of 0.17 ± 0.01, 0.21 ± 0.01, and 0.26 ± 0.20 mas yr⁻¹. At the scale of 7.32 parsecs per mas, this corresponds to a apparent proper motions of 4.00 ± 0.24c, 4.95 ± 0.24c and 6.1 ± 4.7c. The last of these components is only detected in the last three images and has a rather large error bar, but the other two components are secure detections in all four epochs. We re-analyzed the data using a variety of possible scales for the components and found results very consistent with the above.

Zoom In Zoom Out Reset image size

PKS 2209+080 was previously observed and modeled as an IC/CMB X-ray jet by Jorstad & Marscher (2006), where we used their X-ray flux density in our SED. As shown in Figure 8, PKS 2209+080 has a very knotty and straight jet, with the southern hot spot showing two resolved components. Our ALMA and HST observations resolve all of the identified radio knots. As can be seen in both HST images, the host galaxy has an irregular tail structure that is suggestive of a possible recent merger. Only the southern hot spot is detected in the infrared as shown in the lower-middle panel of Figure 8. The only knot reported to show X-ray emission by Jorstad & Marscher (2006) was knot E. In Figure 10, we show the SEDs of knots A through D, which are upstream of knot E, with data shown as squares and synchrotron model fits as solid lines. In knots A through C, there is an upturn in the SED at 600 nm, which lies above a single power-law fit for the synchrotron spectrum. Similar to the cases of PKS 1136–135 and 3C 273, fitting these UV-upturns with an IC/CMB model would require X-ray flux densities much higher than those observed by Chandra.

Zoom In Zoom Out Reset image size

The IC/CMB model for the bright and/or hard large scale jet X-ray emission associated with powerful quasars has now been ruled out for six sources on the basis of over-predicting the observed gamma-ray flux. For each source presented in this paper, we found upper limits to the gamma-ray flux well below the levels predicted by the IC/CMB model using observations from the Fermi/LAT. For the sources not presented in this paper, 3C 273 and PKS 0637–752, previous Fermi/LAT observations showed similar IC/CMB-violating upper limits (Meyer & Georganopoulos 2014; Meyer et al. 2015, 2017). However, over-predicting the observed gamma-ray flux is not the only line of evidence against the IC/CMB model for these sources.

In the case of PKS 1136–135, a high degree of UV polarization with fractional polarization measures in excess of 30% for several knots was detected in the rising second component of the jet SEDs (Cara et al. 2013). This implies a synchrotron origin for the second spectral component of these knots (see Atoyan & Dermer 2004; Harris et al. 2004; Kataoka & Stawarz 2005; Hardcastle 2006; Jester et al. 2006; Uchiyama et al. 2006), as the IC/CMB emission is expected to have very low polarization (see Uchiyama 2008; McNamara et al. 2009) while synchrotron radiation can be highly polarized. These UV-upturns in the knot SEDs are clearly seen in the case of knots A–C for PKS 2209+080 (shown in Figure 10) and knots α and A for PKS 1136–135 (shown in Figure 4). The optical spectrum for these knots is consistently harder than can be fit with a single power law from radio to optical, because the exponential cutoff occurs well before the optical. In these cases, fitting the UV-component of the SED with an IC/CMB model would predict much higher X-ray flux densities than are observed. Furthermore, knots α and A in PKS 1136–135 have a UV spectrum that is much harder than an IC/CMB model fit. These inconsistencies are additional lines of evidence against the IC/CMB model and imply another electron distribution higher in energy to produce the UV-upturns and the X-ray emission.

The apparent misalignment of the sources is also in tension with the IC/CMB model. The IC/CMB model is able to reproduce the observed bright X-ray flux densities by requiring very small jet angles to the line of sight so the X-ray emission is highly relativistically beamed (Tavecchio et al. 2000; Celotti et al. 2001). It has been previously shown by Meyer et al. (2011) that blazars have a higher radio core dominance, R_CE, and lower crossing frequency, ${\nu }_{\mathrm{cross}}$, than radio galaxies. Figure 11, adapted from Meyer et al. (2011), shows the locations of all six sources in the R_CE–${\nu }_{\mathrm{cross}}$ plane in addition to other previously classified blazars and radio galaxies. As can be seen in Figure 11, our sources do not have the low crossing frequencies and high radio core dominances expected for highly aligned blazars. Additionally, PKS 1136–135 and PKS 2209+080 are well within the radio galaxy category, suggesting they are fairly misaligned. It is also apparent from the total SEDs shown in Figure 2 that none of these sources have a Compton dominance greater than 1, which would be expected for highly aligned jets (see Figure 3 from Meyer et al. 2012).

Zoom In Zoom Out Reset image size

One property of the IC/CMB model is that it predicts a different relativistic beaming pattern than synchrotron emission. The expected beaming pattern for IC/CMB radiation is $L=L^{\prime} {\delta }^{p+1+2{\alpha }_{r}}$, with p = 2 for a continuous flow and p = 3 for discrete moving blobs (Dermer 1995; Georganopoulos et al. 2001). Here, L is the luminosity in the galaxy frame assuming isotropy, ${L}^{{\prime} }$ is the solid-angle integrated luminosity in the jet frame, ${\alpha }_{r}$ is the radio spectral index, and δ is the Doppler factor. In the case of synchrotron emission, the expected beaming pattern is given by $L=L^{\prime} {\delta }^{p+{\alpha }_{r}}$ (Dermer 1995).

Because larger angles of the jet to the line of sight decrease the value of δ, the observed IC/CMB flux should fall off faster than the synchrotron flux as source misalignment increases due to the stronger dependence on δ of IC/CMB emission. We define a quantity we refer to as the X-ray dominance, R_x, as the logarithmic ratio of the 1 keV X-ray flux density to the 8.6 GHz radio flux density (in $\nu {F}_{\nu }$). In Figure 12, we show the X-ray dominance plotted against the core dominance for the X-ray brightest knots of the six sources in which IC/CMB is now ruled out as the X-ray emission mechanism, where the core dominance is a measure of the relative misalignment of the jet angle to the line of sight, as can be seen in Figure 11. Using the above relativistic beaming equations for IC/CMB and synchrotron emission, it can be shown that ${R}_{x}\propto {\delta }_{\mathrm{knot}}^{1+{\alpha }_{\mathrm{knot}}}$ and ${R}_{\mathrm{CE}}\propto {\delta }_{\mathrm{core}}^{p+{\alpha }_{\mathrm{core}}}$. Using these equations, we plot the projected misalignment of PKS 0637–752 in Figure 12 as solid black curves, assuming ${\alpha }_{\mathrm{core}}=0$, ${\alpha }_{\mathrm{knot}}=0.7$, and p = 2 for the core, which reflects the most likely case in which the core is a standing shock and not a discrete moving feature. The misalignment curves are plotted for a core Γ of 10 and 50, with the shaded blue region representing the range of misalignment curves for a core Γ between 10 and 50. As can be seen in Figure 12, the other five sources do not fall within this misalignment zone of PKS 0637–752, as would be expected for the stronger beaming of the IC/CMB emission—albeit with a small sample of sources. The lack of correlation between R_x and R_CE suggests the radio-optical synchrotron emission and X-ray emission for these knots have the same relativistic beaming profile, not a different one as would be expected if the X-rays were produced by the IC/CMB mechanism. However, the large range in X-ray dominance also suggests that there is some intrinsic variability of the X-ray component within the sample. A larger sample should allow a more careful examination of the expected correlation, which we leave to future work.

Zoom In Zoom Out Reset image size

All of these sources show what we term a cannonball jet knot structure in the radio with quasi-periodic knot spacing. It has been suggested that this quasi-periodic spacing of the jet knots may be due to re-confinement shocks of the external medium, accretion disk instabilities, or a binary supermassive black hole (Godfrey et al. 2012). Other interesting characteristics of the radio morphology include the bright elongated structure at the end of the jet and lack of extended lobe emission for the sources 3C 273 (Bahcall et al. 1995) and PKS 1136–135 (though there is clearly lobe emission connected to the counter jet in PKS 1136–135), which may be identifiable with the "nose cone" jet structure seen in magnetohydrodynamics simulations. These nose cone structures are thought to occur when the jets are magnetically confined. Rather than the plasma deflecting back from the interface between the jet and intergalactic medium to form the lobes, it is collected in the end of the jet between the Mach disk and leading bow shock, forming the "nose cone." One observational signature associated with the nose cone jets is enhanced radio emission in the nose cone due to an enhanced magnetic field strength and thus synchrotron emissivity, but a decrease in the flux from higher energies due to a lack of particle acceleration in this region from shocks (see Clarke et al. 1986 for a discussion of nose cone jets). This observational characteristic appears to be true in the two cases of 3C 273 and PKS 1136–135, in which we can identify a possible nose cone structure.

The host galaxy of PKS 2209+080, which is visible in our HST observations, shows a clear tidal tail structure that is suggestive of the galaxy having undergone a recent merger. The disturbed elliptical host galaxy of 3C 273 has also shown some evidence of being in a recent merger (Martel et al. 2003). These findings are consistent with the work by Chiaberge et al. (2015), suggesting that all radio-loud AGN have undergone major recent mergers.

One emission mechanism that has not been ruled out for the X-rays in MSC jets is synchrotron radiation from a second higher-energy population of electrons. If these X-ray knots are due to a second electron energy distribution, this means that these jets can exhibit highly efficient in situ particle acceleration very far from the central engine (on the order of kpcs). These electrons would need to be highly energetic (at a few to hundreds of TeV energies), and in cases where the IR/optical flux is totally consistent with a single spectrum from radio to optical, have a lower energy cutoff such that they do not overproduce the observed IR/optical emission. One consequence of these multi-TeV energy electrons is that they should IC scatter the CMB to TeV gamma-rays, which may be detectable with the upcoming CTA for low-redshift sources (Georganopoulos et al. 2006; Meyer et al. 2015).

Alternatively, it is possible these X-rays are due to hadronic emission processes whereby the X-rays are produced by direct proton synchrotron radiation or from the synchrotron radiation produced by secondary electrons resulting from photo-hadronic interactions like the Bethe–Heitler process or photopion production (see Bhattacharyya & Gupta 2016; Kusunose & Takahara 2017; Petropoulou et al. 2017). Like synchrotron models, and unlike IC/CMB, hadronic models are highly tunable and can be made to reproduce most observed SEDs because there are a large number of parameters to adjust. However, one distinguishing characteristic of hadronic models would be the expected neutrino emission from charged pion or neutron decay resulting from interactions between high-energy protons and photons (Mannheim & Biermann 1989). It has been claimed that the bright gamma-ray flare observed in the blazar PKS B1424-418 is hadronic in nature, due to the detection of a PeV-energy neutrino with a high degree of positional and temporal coincidence (Kadler et al. 2016) by the IceCube neutrino detector. However, the detection rate of IceCube high-energy neutrinos is so low that this is not expected to be a very robust means of distinguishing between hadronic and leptonic models for blazars. Additionally, in the case of PKS 0637–752, hadronic emission models require highly super-Eddington jet powers (Kusunose & Takahara 2017).

These models for the X-ray emission do not have many of the problematic features of the IC/CMB model. They do not require these jets to remain highly relativistic on large scales or be highly aligned. The observed "knottiness" of these X-ray jets is well-explained by the much stronger radiative losses for these emission mechanisms. Finally, the second synchrotron model naturally explains the high degrees of polarization measured in the UV-upturns for PKS 1136–135 (Cara et al. 2013) and 3C 273 (Jester et al. 2007).

The empirical fits to the observed synchrotron SEDs are simple power laws with scaled exponential cutoffs, corresponding to a simple power-law electron energy distribution with maximum Lorentz factor. They have the following form:

$$\begin{eqnarray}&&\nu {f}_{\nu }=N{\left(\displaystyle \frac{\nu }{{10}^{10}\mathrm{Hz}}\right)}^{\gamma }\exp \left(-{\left(\displaystyle \frac{\nu }{{\nu }_{1}}\right)}^{\beta }\right).\end{eqnarray} \tag{ 1 }$$

In this equation, ν is the observed frequency of the radiation, γ the power-law index, ${\nu }_{1}$ the frequency at which the exponential turnover begins, β the steepness of the cutoff, and N is the normalization of the spectrum, which has the units erg cm⁻² s⁻¹.

The IC/CMB model has the following form for the shifting in luminosity and frequency of the lower-energy synchrotron spectrum, first derived in Georganopoulos et al. (2006):

$$\begin{eqnarray}&&\displaystyle \frac{{\nu }_{c}}{{\nu }_{s}}=\displaystyle \frac{{\nu }_{\mathrm{CMB}}}{e(B/\delta )/[2\pi {m}_{e}c(1+z)]}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&\displaystyle \frac{{L}_{c}}{{L}_{s}}=\displaystyle \frac{32{U}_{\mathrm{CMB}}{(1+z)}^{4}}{3{(B/\delta )}^{2}},\end{eqnarray} \tag{ 3 }$$

where B is the magnetic field strength, e is the elementary charge, U_CMB is the CMB energy density at the current epoch, ${\nu }_{\mathrm{CMB}}$ the CMB peak frequency at the current epoch, z is the redshift, m_e is the electron mass, c is the speed of light, and δ is the Doppler factor. Note that the only free parameter in this shift is $B/\delta$ which becomes fixed upon fitting the X-ray component of the SED. If one assumes an equipartition magnetic field, this shift is only parametrized by δ. Also note that these shifts should preserve the same spectral index for both the synchrotron and inverse-Compton components, although this is not always seen for these MSC jets.