THE NUCLEAR SPECTROSCOPIC TELESCOPE ARRAY (NuSTAR) HIGH-ENERGY X-RAY MISSION

The two NuSTAR optics modules each contain 133 nested multilayer-coated grazing incidence shells in a conical approximation to a Wolter-I geometry. Table 3 provides the key parameters. Each shell is comprised of either 12 or 24 formed glass segments, depending on the radius in the optic. The glass is thin (0.2 mm) sheet glass manufactured by Schott. The optics are coated with depth-graded multilayer structures, which, for energies above ∼15 keV, increase the graze angle for which significant reflectance can be achieved. This increases the field of view (FoV) and high-energy collecting area relative to standard metal coatings. The coating materials and prescriptions vary as a function of graze angle (or, equivalently, shell radius) and have been designed to optimize the broadband energy response and FoV. The inner 89 shells are coated with depth-graded Pt/C multilayers that reflect efficiently below the Pt K-absorption edge at 78.4 keV. The outer 44 shells are coated with depth-grade W/Si multilayers that reflect efficiently below the W K-absorption edge at 69.5 keV.

Table 2. Key Observatory Performance Parameters

Parameter	Value
Energy range	3–78.4 keV
Angular resolution (HPD)	58''
Angular resolution (FWHM)	18''
FoV (50% resp.) at 10 keV	10'
FoV (50% resp.) at 68 keV	6'
Sensitivity (6–10 keV) (106 s, 3σ, ΔE/E = 0.5)	2 × 10−15 erg cm−2 s−1
Sensitivity (10–30 keV) (106 s, 3σ, ΔE/E = 0.5)	1 × 10−14 erg cm−2 s−1
Background in HPD (10–30 keV)	1.1 × 10−3 counts s−1
Background in HPD (30–60 keV)	8.4 × 10−4 counts s−1
Energy resolution (FWHM)	400 eV at 10 keV, 900 eV at 68 keV
Strong source (>10σ) positioning	15 (1σ)
Temporal resolution	2 μs
Target of opportunity response	<24 hr
Slew rate	006 s−1
Settling time	200 s (typ)

Download table as:  ASCIITypeset image

Compared with soft X-ray telescopes, the optics prescription utilizes smaller graze angles (the angle between an on-axis incident X-ray and the optics shell). This means vignetting losses will become noticeable for off-axis angles above 2'. This factor, when combined with the dependence of the multilayer reflectivity on both the incident angle and energy of the X-ray (higher energy photons have higher reflectivity at shallower graze angles), results in a complex behavior for the effective area of the optic. As illustrated in Figure 4, the optics area as a function of off-axis source position decreases with increasing photon energy. The effective FoV, defined by the furthest off-axis position that has 50% of the on-axis effective area, will therefore decrease with energy: it is 10' at 10 keV and 6' at 68 keV.

Zoom In Zoom Out Reset image size

The optics point-spread function (PSF) is dominated by figure errors inherent in the substrates and in the mounting technique. As a result, unlike Chandra, where the response is dictated by the optical design, for NuSTAR the PSF is not a strong function of off-axis angle. While the detailed shape changes, to first order the area of the encircled energy contours remains approximately constant with off-axis angle.

A detailed description of the optics can be found in Hailey et al. (2010), the coatings are described in Christensen et al. (2011), the substrate production is described in Zhang (2009), and the overall optics fabrication approach is detailed in Craig et al. (2011).

Each telescope has its own focal plane module, consisting of a solid state CdZnTe pixel detector (Harrison et al. 2010) surrounded by a CsI anti-coincidence shield. Table 4 provides the key focal plane parameters. A two-by-two array of detectors (Figure 5), each with an array of 32 × 32, 0.6 mm pixels (each pixel subtending 123), provides a 12' FoV. In NuSTAR, each pixel has an independent discriminator and individual X-ray interactions trigger the readout process. On-board processors, one for each telescope, identify the row and column with the largest pulse height and read out pulse height information from this pixel as well as its eight neighbors. The event time is recorded to an accuracy of 2 μs relative to the on-board clock. The event location, energy, and depth of interaction in the detector are computed from the nine-pixel signals (see Rana et al. 2009; Kitaguchi et al. 2011 for details). The depth-of-interaction measurements allow an energy-dependent cut to be made to reduce internal detector background.

Zoom In Zoom Out Reset image size

Table 4. Focal Plane Parameters

Focal Plane Parameter	Value	Focal Plane Parameter	Value
Pixel size	0.6 mm/123	Max. processing rate	400 events s−1 module−1
Focal plane size	12' × 12'	Max. flux meas. rate	104 counts s−1
Hybrid format	32 pix × 32 pix	Time resolution (relative)	2 μs
Energy threshold	2 keV	Dead time fraction (at threshold)	5%

Download table as:  ASCIITypeset image

Since the detectors do not employ an integrating CCD-style readout, pulse pileup will not occur until source fluxes of ∼10⁵ counts s⁻¹ pixel⁻¹. The processing time per event is 2.5 ms, limiting the rate at which events can be read out to between 300 and 400 events s⁻¹ module⁻¹. For each event, the live time since the previous event is recorded to an accuracy of 1 μs, so that fluxes can be measured to 1% accuracy even for incident count rates of 10⁴ cps. Harrison et al. (2010) give a detailed description of the operation of the custom ASIC readout.

The focal plane detectors are placed inside a CsI anti-coincidence shield. Events resulting in a simultaneous energy deposition in both the surrounding anti-coincidence shield and the detector are rejected on board by a processor as background. The opening angle of the shield is large (15° FWZI), greatly exceeding the sky solid angle blocked by the optics bench; thus, to limit diffuse background impinging on the detector, a series of aperture stops were deployed post-launch simultaneously with the mast. The aperture stops collimate the detector to a 4° FWZI field, which is partially but not completely blocked by the optics bench. This results in some stray light, which dominates the background below 10 keV (Section 5.3).

The NuSTAR mast structure provides the required 10.14 m separation between the optics and the focal plane detectors. The mast was engineered by ATK Space Systems from segments made up of carbon fiber, aluminum, and steel components, designed to be as isothermal as possible in order to minimize structural distortions. The 57 bays (see Figure 3) of the mast were deployed on orbit and locked in place by tensioned steel cables. Although the mast has a low net coefficient of thermal expansion, orbital day/night extremes create residual deflections that result in motion of the optical axis and the X-ray focal point on the focal plane detectors. Thermal modeling of the system pre-launch indicated that the deflection would range from less than 1 mm per orbit in the most favorable orientations to a few mm in the least favorable.

The mast motion is tracked by the combination of a metrology system (Liebe et al. 2012) and the instrument star camera head. The metrology system consists of two IR lasers mounted on the optics bench with their beams focused on two corresponding detectors on the focal plane bench. The lasers are mounted in a temperature-controlled Invar structure rigidly mounted to the optics bench. The star tracker is also mounted to that bench, which was engineered from carbon fiber structural elements to provide less than 2'' of relative deflection with respect to the optics modules during on-orbit thermal conditions.

The laser beam positions are monitored by silicon position-sensitive detectors mounted near the focal plane modules. Filters and baffles mounted at the detectors reject scattered Sun and Earth light and provide for high signal to noise, producing ∼10 μm (01) accuracy for each laser spot position. On-ground analysis of signals from the two detectors, combined with the star tracker quaternion, separates translational and rotational deflections in the positions of the two benches and enables both accurate correction of the position of the X-ray photons on the focal plane modules and construction of the vignetting function appropriate for a given observation.

The PSF of the telescopes is a convolution of the optics and focal plane detector response combined with residual errors from the metrology system and attitude reconstruction. The dominant contributor is the optics, with residual terms arising from finite detector sampling and imperfect mast motion and attitude reconstruction at the few-arcsecond level. The PSF has been measured in-flight using high signal-to-noise observations of bright point sources. The PSF shape consists of a sharp inner core with a full width at half-maximum (FWHM) of 18'' and a half-power diameter (diameter of a circle enclosing half of the X-ray counts from a point source) of 58''. Figure 6 shows the encircled counts fraction as a function of diameter for a circular extraction region (left panel) as measured for telescope module A (the telescope and focal plane modules are designated by A and B), from an observation of the accreting Galactic binary black hole GRS 1915+105. The right panel of the figure shows a stretched image, demonstrating the azimuthal symmetry of the response. The PSF for module B is within 2'' of that of module A.

Zoom In Zoom Out Reset image size

The normalization of the NuSTAR effective area in the 2–10 keV band has been verified in flight by simultaneous observations of the Crab Nebula and its pulsar, and the quasar 3C 273 with XMM-Newton, Suzaku, Swift, and Chandra. Flux measurements in the 5–10 keV band derived for NuSTAR have been compared for the indicated calibration observations, as well as for science targets. We find that the overall normalization in the 5–10 keV band agrees well with the XMM-Newton PN, the Suzaku XIS, and Swift with typical cross-calibration corrections at the 10% level, similar to the cross-calibration factors among the various soft X-ray observatories. The NuSTAR relative response also matches nicely with XMM-Newton PN and MOS, and Suzaku XIS in the 3–10 keV band (see Figure 7). In a joint observation of the active galaxy IC 4329A with NuSTAR and Suzaku, the absolute flux in the 3–10 keV band measured with XIS 0 was (8.30 ± 0.03) × 10⁻¹¹ erg cm⁻² s⁻¹, and for focal plane modules A and B the measured flux was (9.23 ± 0.03) × 10⁻¹¹ erg cm⁻² s⁻¹ and (8.95 ± 0.03) × 10⁻¹¹ erg cm⁻² s⁻¹, where quoted errors represent 90% confidence intervals.

Zoom In Zoom Out Reset image size

The systematic calibration errors between NuSTAR's A and B telescopes are at the 3% level, with a 10% normalization uncertainty relative to Suzaku. Direct comparisons with INTEGRAL-ISGRI and Suzaku-PIN at higher energies are more difficult, with larger cross-normalization factors and higher uncertainty in the case of the PIN due to challenges with background subtraction. However, these cross-calibrations are ongoing, and will be aided by high count-rate joint observations of Cygnus X-1 and the Crab.

The in-flight spectral response of the NuSTAR focal plane detectors matches well with pre-launch calibration measurements. Each focal plane module is equipped with a radioactive ¹⁵⁵Eu calibration source that can be placed in the FoV. Figure 8 shows a spectrum taken during on-orbit commissioning. At energies below ∼50 keV, the resolution is 400 eV FWHM and is constant with energy, being determined by the electronic noise. Above this, the energy resolution broadens due to charge trapping effects in the detector. The FWHM response is 1.0 keV at the 86 keV decay line. The good resolution over a broad energy range enables NuSTAR to perform spectroscopy extending from below the Fe-line range up to 79 keV. Figure 9 shows a spectrum from a 15 ks exposure of the accreting neutron star Vela X-1 taken by NuSTAR during the instrument commissioning phase. Fe lines are well measured, along with the continuum extending up to 79 keV. A clear cyclotron resonance scattering feature can be seen at 55 keV.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Because NuSTAR has a triggered readout (similar to a proportional counter and unlike an integrating CCD), timing analysis is possible on short timescales for sources of moderate count rate (<200 counts module⁻¹). For very bright targets, the 2.5 ms deadtime per event limits the ability to search for features on millisecond timescales. NuSTAR maintains relative stability of event timing to 1–2 ms by correcting the drift of the spacecraft clock on long (day to weeks) timescales. This drift correction is done using the relative spacecraft clock time to UT that is routinely measured during ground station contacts. The absolute time relative to UT is currently only known to 5 ms; however, calibration to the sub-millisecond level using a pulsar with a known ephemeris is planned.

The NuSTAR background measured on orbit is well within the range predicted by pre-launch Monte Carlo modeling. Background prediction for a well-shielded instrument is difficult due to the poorly characterized input particle spectra, so that pre-launch predictions had uncertainties of a factor of four. Figure 10 shows the background spectrum taken during an observation of the Extended Chandra Deep Field South (ECDFS) in late 2012 September. At low energies (E < 10 keV), the background is dominated by diffuse cosmic flux entering through the aperture stop. This component is not uniform across the FoV, so some care is required in choosing regions for background subtraction at low energies for faint sources. The internal background consists of albedo from Earth's atmosphere and fluorescence lines from the CsI shield, as well as internal activation lines produced largely by particles trapped in the SAA. This component is spatially uniform across a given detector chip, but does vary from detector to detector.

Zoom In Zoom Out Reset image size

The low inclination of the NuSTAR orbit results in stable background compared with missions that experience a larger range of geomagnetic latitude and more frequent SAA passages. The background increased slowly in the several months after launch as a result of activation components, although it has now largely stabilized. On orbit, the background in the 3–79 keV band varies primarily as a function of geomagnetic latitude, with the exception of the interval from 24 to 26 keV, where two lines with half-lives of less than 1 hr result in variations dependent on time elapsed since an SAA passage. For observations significantly longer than an orbit, these variations average out, leaving systematic uncertainties in background subtraction of <1%.

Two modes of background subtraction have been developed depending on whether or not a source-free background region is available in the FoV. For point sources with background regions available on the same chip, standard procedures employed in soft X-ray focusing telescopes apply. For extended sources where clear background extraction regions do not exist, the background can be scaled from deep survey fields and filtered to match the average geomagnetic cutoff for the observation. Both techniques work well and can be applied to extract spectra at levels well below that of the background.

The NuSTAR mast, optics bench, and focal plane bench were co-aligned on the ground to within a few millimeters. This ensured that on orbit the deployed benches would be well within the adjustment capability of the mast adjustment mechanism mounted at the tip of the mast. After deployment on orbit, the mast tip was adjusted in two steps to place the optical axis of the telescopes at the desired position on the focal plane detectors. X-ray sources with known positions were then observed to measure the relative positions of the optics, focal plane detectors, laser metrology system, and optics-bench-mounted star tracker.

As described in Section 4.3, the metrology system measures the relative positions of the optics and focal plane benches while the star tracker provides absolute celestial reference. Due to thermal effects, the benches move relative to one another in translation over an orbit that, when combined with the mast tilt, results in motion of the optical axis on the X-ray focal plane detectors by 3–4 mm (1'–12; see Figure 11). This relative motion is measured and removed using the metrology system. The absolute orientation of the optics bench is measured by the star tracker.

Zoom In Zoom Out Reset image size

For absolute alignment, we used a set of reference observations to determine the relative alignment of the instrument components. The absolute astrometric positions of bright X-ray sources with known positions were observed to be within ±8'' (90%) using the updated in-orbit alignment data. The systematic astrometry offsets are a function of aspect relative to the Sun, as well as seasonal variations affecting the average illumination during non-occulted viewing periods. Further updates to the alignment parameters that take longer term variations into account are expected to result in a final performance (systematic contribution) of ±3'' (90%), with the error being dominated by the inherent stability and accuracy of the optics-bench-mounted star tracker.

Resolving the sources that make up the cosmic X-ray background (XRB) has been a primary driver for X-ray astronomy over the last 50 yr. Huge strides have been made with sensitive surveys undertaken by the Chandra and XMM-Newton observatories (e.g., Alexander et al. 2003; Harrison et al. 2003; Luo et al. 2008; Elvis et al. 2009; Laird et al. 2009). These surveys have resolved 70%–90% of the 0.5–10 keV XRB (e.g., Worsley et al. 2005; Hickox & Markevitch 2006; Xue et al. 2011), revealing obscured and unobscured AGNs out to z ∼ 5–6 (e.g., Tozzi et al. 2006; Brusa et al. 2009; see Brandt & Alexander 2010 for a recent review). However, Chandra and XMM-Newton are only sensitive to sources emitting below 10 keV, far from the ∼20–30 keV peak and therefore leaving significant uncertainties on the properties of the sources that dominate (e.g., Ajello et al. 2008; Ballantyne et al. 2011). By contrast, the directly resolved fraction of the XRB at the 20–30 keV peak from current >10 keV surveys is only ∼2% (e.g., Krivonos et al. 2007; Ajello et al. 2008), and while the sources detected in Chandra and XMM-Newton surveys must resolve at least an order of magnitude more, no direct observational constraints are currently available.

The NuSTAR extragalactic survey will provide a significant breakthrough in revealing the composition of the XRB at the 20–30 keV peak, achieving individual source detection limits approximately two orders of magnitude fainter than those previously obtained at >10 keV (e.g., Bird et al. 2010; Tueller et al. 2005; Ajello et al. 2012). This will provide a clean obscuration-independent selection of AGN activity (at least out to N_H ≈ 10²⁴ cm⁻²), required to test claims from Chandra and XMM-Newton surveys that the fraction of obscured AGNs increases with redshift (e.g., La Franca et al. 2008; Treister & Urry 2006). The NuSTAR extragalactic survey (Table 7) has three primary components: (1) a deep (∼200–800 ks) small-area (∼0.3 deg²) survey of the ECDFS, (2) a medium (∼50–100 ks) wider-area (1–2 deg²) survey of the COSMOS field, and (3) a shallow (∼15 ks) serendipitous survey (∼3 deg²), achieved by targeting 100 Swift-BAT AGNs. With this multilayered approach, the NuSTAR extragalactic survey will cover a broad range of the flux-solid angle plane and is expected to directly resolve 30%–50% of the 10–30 keV background from individual source detections alone (e.g., Ballantyne et al. 2011). Sensitive average constraints on the composition of the XRB will also be obtained by stacking the NuSTAR data of the faint X-ray sources detected at E < 10 keV by Chandra and XMM-Newton in the ECDFS and COSMOS fields.

Figure 12 shows the expected luminosity–redshift distribution for the different NuSTAR extragalactic surveys, derived using the methodology in Ballantyne et al. (2011), along with the positions in this diagram for the first seven NuSTAR serendipitously detected AGNs (D. M. Alexander et al. 2013, in preparation). We predict ∼160–200 NuSTAR-detected sources (∼65–80 at 10–30 keV) from the medium and deep NuSTAR survey components, the majority at redshifts between 0.5 and 2 with X-ray luminosities in the range of L_X ∼ 10⁴³–10⁴⁵ erg s⁻¹. These ranges in redshift and luminosity correspond to the peak epoch in the black hole accretion history of the universe and the knee of the X-ray luminosity function (e.g., Ueda et al. 2003; La Franca et al. 2008; Aird et al. 2010); thus, NuSTAR should detect the AGN source population that dominates the cosmic growth of black holes. The shallow serendipitious survey will achieve two principal aims: (1) provide the identification of serendipitous sources to trace the bright end of the >10 keV XRB population and (2) provide the best global characterization of the high-energy component of the nearby AGN population (e.g., absorption, reflection strength, and high-energy cutoff), to allow for the more accurate modeling of the XRB source populations. We predict ∼100 NuSTAR-detected sources in the serendipitous survey (∼30 at 10–30 keV), the majority at z < 1 with L_X ∼ 10⁴³–10⁴⁵ erg s⁻¹.

Zoom In Zoom Out Reset image size

NuSTAR's Galactic surveys will extend hard X-ray studies to fainter levels with the aim of detecting large numbers of compact objects (black holes, neutron stars, and white dwarfs), determining the hard X-ray morphology of the diffuse emission in the central few dozen parsecs of the Galaxy, monitoring the temporal and spectral properties of the supermassive black hole (SMBH) Sgr A*, and determining the origin of the Galactic diffuse XRB.

The Galactic center and Galactic bulge contain large numbers of compact objects (X-ray binaries and cataclysmic variables, CVs) of interest for accretion studies, as well as highly magnetized or rapidly rotating neutron stars (NS). By detecting compact stellar remnants in hard X-rays, we can elucidate how populations evolve from groups of single and binary high-mass and low-mass stars to reach the ends of their lives as compact objects. Chandra has previously uncovered in excess of 6000 sources in a 2° × 08 region (Muno et al. 2009). There is evidence that CVs are the largest component of this population. However their nature, whether magnetic or non-magnetic, and their classification remain uncertain. Determining hard X-ray spectral and timing properties will provide important constraints.

NuSTAR will perform a deep survey of a 2° × 08 region around the Galactic center and, depending on spectral hardness of the source population, should detect up to 100 compact objects. In addition to the CVs, there are also 30 X-ray sources that have been identified as massive stars (Mauerhan et al. 2010), indicating the presence of a significant younger population. INTEGRAL observations of this region are extremely source-confused, and NuSTAR's vastly better PSF is required to determine which sources have hard X-ray counterparts. It is also unclear from the Chandra observations whether the massive stars are single or binary and whether any binaries harbor compact objects. There are currently only about 100 high-mass X-ray binaries (HMXBs) known in the entire Galaxy, and NuSTAR (via hard X-ray spectral and timing properties) can determine whether more HMXBs are present in this region.

NuSTAR will also look at other regions of the Galaxy with populations of different ages. A major open question in Galactic dynamics is the relative evolutionary states of the different spiral arms. It is thought that the Carina arm (Smith & Brooks 2007) is much less evolved than the Norma arm, which, in turn, is less evolved than the Galactic center region; however, this has not been demonstrated unambiguously. While the number of known HMXBs has increased with INTEGRAL, it is still likely that the HMXB population is being significantly undercounted. If it is true that the Norma region population is significantly younger than the Galactic center population, then one would expect NuSTAR to find a much higher HMXB-to-CV fraction.

NuSTAR will survey the 2° × 08 region in Norma covered by Chandra (Muno et al. 2009). While Chandra provides subarcsecond source positions and soft X-ray information, a NuSTAR survey of the same region will expand our understanding of the source populations. For example, INTEGRAL results suggest that there may be a large number of low-luminosity HMXBs, an important consideration when the HMXB luminosity function is used as a star formation rate indicator in distant galaxies (Grimm et al. 2003). Furthermore, understanding HMXB evolution (as well as the overall number of HMXBs) will help constrain the relative numbers of NS/NS, NS/BH, and BH/BH binaries (Tauris & van den Heuvel 2006).

The Galactic surveys are also elucidating the morphology of the central few tens of parsecs around the Galactic center. This region contains X-ray emission from the SMBH Sgr A*, numerous pulsar wind nebulae, supernova remnants (SNRs), and molecular clouds (Baganoff et al. 2003). Of particular interest for NuSTAR are the flares from Sgr A*. These flares, seen about once per day, have been detected from radio wavelengths all the way to the soft X-ray band, but only hard X-ray upper limits existed until NuSTAR observed a large flare in 2012 July. The NuSTAR data, in conjunction with observations at lower energies, put tight constraints on emission mechanisms (Barrière et al. 2013). Further campaigns are planned for next year, to be coordinated with Chandra, XMM, and Keck. INTEGRAL has observed this region extensively (Bélanger et al. 2006); however, it is severely source-confused given the 12' FWHM PSF. With subarcminute angular resolution and ∼3'' localizations, NuSTAR is searching for hard X-ray sources and mapping the complex diffuse emission surrounding Sgr A* out to ∼50 keV.

NuSTAR will also map diffuse hard X-ray emission from molecular clouds. In particular, Sgr B2 has recently attracted much attention, since its diffuse Fe–K and continuum X-ray emission are time correlated, as would be expected if the emission is in response to intense flaring activity of some bright source, possibly Sgr A*, in the past (Sunyaev & Churazov 1998). A less likely alternative is injection of cosmic rays into the cloud (Yusef-Zadeh et al. 2007) with subsequent X-ray emission. These processes predict distinct morphological differences in the hard X-ray continuum emission, enabling us to distinguish between the two models; furthermore, by measuring the strength of the inverse Compton continuum emission, we can constrain the density of the cloud and therefore the required input energy (Ponti et al. 2012).

The 2° × 08 Sgr A* survey when combined with the observation of the low column density Limiting Window, NuSTAR should also be able to resolve the nature of the Galactic diffuse high-energy XRBs. Likely due to unresolved point sources (Revnivtsev et al. 2009; Hong 2012), it is not clear what fraction is due to magnetic CVs versus coronally active giants, which would have relatively soft spectra. NuSTAR's ability to measure hard X-ray spectral properties should settle this issue.

Many active galaxies possess prominent relativistic jets, and if they point close to our line of sight, jet emission can dominate the radiative output of the object from radio to very energetic γ-rays. Termed "blazars", the emission from these sources is highly variable, by up to factors of 100 in the optical (see, e.g., Wagner & Witzel 1995) and 1000 in γ-ray (see, e.g., Ackermann et al. 2010; Wehrle et al. 2012). Blazar spectral energy distributions generally consist of two prominent, broad humps, one peaking in the millimeter-to-X-ray spectral range and the other in the hard X-ray-to-high-energy γ-ray range. Synchrotron emission is responsible for the low-energy peak, while inverse Compton produces the gamma-ray hump. General coincidence of variability patterns in the far-IR to optical band with the γ-ray band suggests that the two processes are likely to originate from the same population of highly relativistic electrons, with Lorentz factors γ_el reaching 10⁶ or higher. The situation in the X-ray band is less clear, since the variability is only modestly correlated with that in the optical and γ-ray bands (see, e.g., Bonning et al. 2009; Hayashida et al. 2012).

Hard X-ray measurements are important diagnostics of the content of the jet, as well as of the processes responsible for the acceleration of particles to the very highest energies. In luminous blazars—often associated with powerful quasars—the X-ray spectra are generally hard, suggesting that the X-ray band represents the onset of the inverse Compton component (see, e.g., Kubo et al. 1998). The low-energy end of the Compton component is produced by particles with the highest energies, since the particle spectrum falls steeply with energy so that X-ray observations constrain the total particle content of the jet (see, e.g., Sikora & Madejski 2000). Conversely, in the low-luminosity objects, the X-ray band is thought to be the high-energy end of the synchrotron component, and is therefore produced by the most energetic "tail" of the electron spectrum (cf. Takahashi et al. 1996). Hard X-ray observations conducted jointly with the TeV γ-ray observations probe electrons with most extreme energies. This is especially important for the most energetic TeV emitters, which reach the highest apparent values of γ_el, since they become radiatively inefficient in the Klein–Nishina regime. Measuring the most energetic synchrotron emission allows one to distinguish between intrinsic and extrinsic cutoff mechanisms, and sets clear constraints on γ_el. NuSTAR's large bandwidth provides a unique tool for such investigations.

The extreme variability of blazars provides an opportunity to study the evolution of the radiating particles' energy distribution, from the acceleration stage through their luminous emission. Broadband studies can probe the conversion of jet bulk kinetic energy to ultrarelativistic electrons and subsequently to photons. The relative temporal variability in different spectral bands, from radio through TeV, provides an important handle on the location of the energy dissipation as a function of distance from the central black hole (Sikora et al. 2009).

As the main component of its blazar program, NuSTAR will participate in large multiwavelength campaigns of two luminous blazars showing quasar-like behavior, and two low-luminosity sources in an active state. Extreme activity is unpredictable (see, e.g., Wagner & Witzel 1995; Abdo et al. 2010), so that targets must be chosen by monitoring activity in the Fermi band. In the quasar class, 3C279 (z = 0.524) and 3C454.3 (z = 0.859) are the most promising targets. Both have shown extreme variability in Fermi on a timescale of ∼3 days (Abdo et al. 2010). NuSTAR will observe with a daily cadence of 10 ks pointings with facilities from radio to γ-ray with sampling as simultaneously as possible.

The best candidates for low-luminosity blazars are Mkn 421 and PKS 2155−304. The most extreme variability for these is seen in the TeV band, on timescales of an hour or less (see, e.g., Aharonian et al. 2009; Galante 2011). Assuming these stay active, NuSTAR will observe simultaneously with MAGIC and Veritas (for Mkn 421), and H.E.S.S. (for PKS 2155−304) for several nights each month during the TeV telescopes' observing seasons. The continuous pointings will be strictly simultaneous (modulo Earth occultations). Preliminary data obtained for Mkn 421 during the NuSTAR calibration phase already indicate that such measurements of time-resolved spectra with NuSTAR are feasible, where a ∼15 ks observation demonstrably samples the spectrum to 50 keV.

NuSTAR will also make shorter observations of several blazars with particularly unusual properties. Candidates include 1ES 0229+200, where hard X-ray and TeV γ-ray emission have interesting implications for the intergalactic magnetic field (see Kaufmann et al. 2011; Vovk et al. 2012), and AO 0235+164 (see Ackermann et al. 2012; Agudo et al. 2011). The scheduling of these targets will depend on them entering an active state as determined by Fermi.

Young (τ ≲ 1000 yr old) SNRs are priority targets for NuSTAR, which will provide the first sensitive imaging observations of non-thermal emission above 10 keV. X-ray synchrotron emission is known to extend to high energies in several young remnants (see Reynolds 2008; Vink 2012 for reviews) and possibly in others; its characterization addresses questions of particle acceleration and the origin of cosmic rays (e.g., Reynolds 1998; Warren et al. 2005). Remnants younger than 1000 yr may also show evidence of relatively long-lived radioactive nuclides, primarily ⁴⁴Ti, which is of importance in understanding explosion mechanisms (Arnett et al. 1989; Timmes et al. 1996; Magkotsios et al. 2010).

The primary remnants for these studies are Cas A, SN 1987A, and G1.9+03—objects where ⁴⁴Ti emission has been detected, and which are also interesting for studying the hard synchrotron continuum. ⁴⁴Ti (mean life 85 yr) decays by electron capture to ⁴⁴Sc, leaving an inner-shell vacancy which can be filled by a transition producing a 4.1 keV X-ray. Nuclear de-excitation lines at 68 and 78 keV and 1.157 MeV are also produced. So far, the gamma-ray lines have been seen only from Cas A, by BeppoSAX (Vink et al. 2001) and IBIS/ISGRI on INTEGRAL (Renaud et al. 2006), and recently from SN 1987A by IBIS/ISGRI (Grebenev et al. 2012). The ⁴⁴Sc X-ray line has been seen only in the youngest Galactic remnant G1.9+0.3 (Borkowski et al. 2010), whose age is about 100 yr (Carlton et al. 2011).

Theoretical predictions for the initial synthesized mass of ⁴⁴Ti in core-collapse SNe are in the range of (0.1–3) × 10⁻⁴ M_☉ (Timmes et al. 1996; Thielemann et al. 1990; Woosley & Hoffman 1991; Young et al. 2006). In Cas A, the inferred yield is about 2 × 10⁻⁴ M_☉ of ⁴⁴Ti, at the upper end of the range of expectations. The yield in SN 1987A is also surprisingly high, (3.1 ± 0.8) × 10⁻⁴ M_☉ (Grebenev et al. 2012). NuSTAR is executing several long observations of Cas A, with over 1.2 Ms of total exposure will make the first spatial maps of the remnant in ⁴⁴Ti. The asymmetry of the emitting region will provide strong constraints on the asymmetry of the explosion. Similarly in SN 1987A, assuming the INTEGRAL detection is robust, NuSTAR's 1.2 Ms integration will significantly improve the yield measurement and will constrain the line width as well as search for hard X-ray emission from any compact object.

The very high expansion velocities detected spectroscopically from G1.9+0.3 (>12,000 km s⁻¹), a century after the explosion, argue fairly strongly for a Type Ia origin, supporting arguments derived from the remnant morphology and the absence of a pulsar wind nebula (PWN; Reynolds 2008). However, the SN-type question is still open. The ⁴⁴Sc 4.1 keV line intensity detected from G1.9+0.3 (Borkowski et al. 2010) implies about 10⁻⁵ M_☉ of ⁴⁴Ti (for an assumed age of 100 yr). The predicted fluxes in the 68 and 78 keV lines are about 4 × 10⁻⁶ ph cm⁻² s⁻¹ each. A detection of ⁴⁴Ti from a firmly classified Type Ia remnant will be a very important constraint on explosion models. If G1.9+0.3 turns out to be a core-collapse remnant, having another example alongside Cas A and SN 1987A will be important in isolating the important variables determining ⁴⁴Ti production.

All Galactic SNRs less than 2000 yr old, and some older examples, show evidence of synchrotron X-ray emission (Reynolds 2008; Allen et al. 2001). In all observed cases, the intensity is below the extrapolation of the radio synchrotron spectrum (Reynolds & Keohane 1999), indicating that some processes (radiative losses, finite remnant age or size, or particle escape) are beginning to cut off the electron acceleration process at high energies. Considerably more information on particle acceleration is encoded in the detailed shape of the spectral steepening than in the power-law portion of the spectrum at lower energies. Most inferences to date are based on observations below about 8 keV, where power laws are indistinguishable from more detailed models. Both Cas A and G1.9+0.3 have strong X-ray synchrotron components; in the case of Cas A, spatially integrated observations to 80 keV have been performed (Favata et al. 1997; Allen et al. 2001; Maeda et al. 2009). NuSTAR will provide spatially resolved maps. The integrated spectrum of G1.9+0.3 is dominated by non-thermal emission below about 7 keV (Reynolds 2008), but no observations have been done at higher energies. Since G1.9+0.3 is the only Galactic SNR currently growing brighter (Carlton et al. 2011), the behavior of its spectrum at higher energies is of great interest. For SN 1987A, X-ray emission below 10 keV is composed of two thermal components (e.g., Sturm et al. 2010), but our long observation to search for 68 keV line emission should be able to detect or constrain any non-thermal continuum at higher energies.

NuSTAR will provide new insights into SN explosions in our local universe by studying the hard X-ray emission from SN Ia explosions out to the Virgo cluster (∼14 Mpc) and core-collapse SNe in the Local Group (∼4 Mpc). This emission arises from the Compton downscatter of ⁵⁶Ni decay gamma rays, and both the light curve and spectrum emerging from the SNe at hard X-rays is sensitive to the mass and abundance distributions in the explosion. Detailed simulations of SNe Ia have shown that this hard X-ray emission is a strong diagnostic of the surface composition and flame propagation, and larger ⁵⁶Ni mass produces higher hard X-ray flux levels that peak at earlier times (Maeda et al. 2012). Typical peaking times are expected to be between 10 and 30 days for SNe Ia and between 150 and 250 days for core-collapse SNe. Our primary criteria for triggering an SN ToO is to have a spectroscopically confirmed SN identified before peak within the NuSTAR sensitivity distances.

NuSTAR will pursue a range of science objectives significantly broader than the key projects described above. The sections below detail additional science objectives and describe which targets will be observed during the baseline mission. Fully exploiting the range of opportunities will require a GI program, to be proposed to NASA as an extension to the baseline mission. Approximately 70% of the total observing time in the baseline mission is allocated to the Priority A targets plus the key (Level 1) programs; the remaining time (30%) is kept in reserve for followup of unanticipated discoveries, as well as additional ToO observations.

10.6.1. Ultraluminous X-Ray Sources

10.6.2. Accretion Physics in Active Galactic Nuclei

10.6.3. Galaxy Clusters and Radio Relics

10.6.4. Nearby Star-forming Galaxies

10.6.5. Magnetars and Rotation-powered Pulsars

10.6.6. Pulsar Wind Nebulae

10.6.7. Obscured AGNs

10.6.8. Galactic Binaries

10.6.9. Solar Physics