Cross-correlating the patchy screening and kinetic Sunyaev-Zel’dovich effects as a new probe of reionization Darby M. Kramer, Alexander van Engelen, Christopher Cain, Niall MacCrann, Hy Trac, Skylar Grayson, Evan Scannapieco, Blake Sherwin Abstract The kinetic Sunyaev-Zel’dovich effect (kSZ) and patchy screening effect are two complementary cosmic microwave background (CMB) probes of the reionization era. The kSZ effect is a relatively strong signal, but is difficult to disentangle from other sources of temperature anisotropy, whereas patchy screening is weaker but can be reconstructed using the cleaner polarization channel. Here, we explore the potential of using upcoming CMB surveys to correlate a reconstructed map of patchy screening with (the square of) the kSZ map, and what a detection of this cross-correlation would mean for reionization science. To do this, we use simulations and theory to quantify the contributions to this signal from different redshifts. We then use the expected survey properties for CMB-S4 and CMB-HD to make detection forecasts. We find that, for or our fiducial reionization scenario, CMB-S4 will obtain a hint of this signal at up to 1.8 ​ σ 1.8\sigma , and CMB-HD will detect it at up to 14 ​ σ 14\sigma . We explore the physical interpretation of the signal and find that it is uniquely sensitive to the first half of reionization and to the bispectrum of the ionized gas distribution. I Introduction The epoch of reionization (EoR) is a pivotal but elusive time in the history of the Universe. It is the period during which the population of hydrogen went from being neutral to ionized, likely due to the formation of the first stars and galaxies [ 74 , 69 , 73 ] . These first objects are difficult to measure directly because of their faint and highly redshifted nature, and because of the structures that continued to form after the EoR, such as larger, brighter galaxies and galaxy clusters that emit at a variety of wavelengths and intensities. Since reionization was caused by individual objects, it is expected to be “patchy,” causing small, ionized bubbles to form and grow over time [ 29 , 72 , 23 , 17 , 62 ] . Many experiments and studies seek to extract signals that solely trace the EoR, such as probes of the galaxy population [ 3 , e.g.,] , mean electron optical depth [ 5 , Planck :] , neutral hydrogen density as traced by 21cm emission [ 61 , e.g.] and Lyman-alpha absorption [ 13 ] at high redshifts. Each of these probes use unique tools and techniques to access the EoR. A particular cosmological tool that can benefit EoR studies in the coming years is the cosmic microwave background (CMB). When the primordial plasma expanded and cooled enough for electrons and protons to combine and form the first atoms at redshift z\sim 1100 , photons were able to free-stream in all directions as a blackbody background, the CMB, seen today with temperature T_{0}=2.7 K. However, there were small density fluctuations traceable by the CMB temperature anisotropies that grew over time to become the large-scale structures we observe today. These temperature fluctuations are known as primary CMB anisotropies and are on the order of 100 \mu K today. Furthermore, some of the CMB photons have interacted with the large scale structure that formed in the later Universe, causing secondary CMB anisotropies. Spectrum Type Reionization Probe Example Reference(s) Forecasted “S4” SNR Range Auto-spectra \langle T_{\mathrm{kSZ}}\times T_{\mathrm{kSZ}}\rangle Raghunathan and Omori [64] \sim 70-80\sigma \langle K\times K\rangle Smith and Ferraro [78] \sim 100\sigma \langle\tau\times\tau\rangle Dvorkin and Smith [22] , Jain et al. [37] \sim 2-3\sigma \langle y\times y\rangle Sun et al. [81] \sim 3\sigma Cross-spectra \langle\tau\times T_{21}\rangle Meerburg et al. [51] \sim 10\sigma \langle K\times T_{21}\rangle Ma et al. [45] \sim 50\sigma \langle\tau\times\kappa_{\mathrm{CMB}}\rangle Feng and Holder [24] , Bianchini and Millea [10] \sim 10-22\sigma \langle\tau\times y\rangle Namikawa et al. [54] , Remazeilles et al. [67] \sim 2-7\sigma \langle K\times\delta_{\mathrm{g}}\rangle La Plante et al. [41] \sim 4-11\sigma \langle y\times\delta_{\mathrm{g}}\rangle Baxter et al. [8] <2\sigma Table 1: Review of some previously-forecasted probes of reionization with CMB data and their reported signal-to-noise ratios. Above the thick horizontal line are the three auto-power spectra discussed in Section I that contain reionization information. Below this line are a few cross-correlations that have been studied in the literature attempting to isolate the reionization information in their constituents. The CMB-S4 (or S4-like) signal-to-noise forecasts for these probes are in the rightmost column. A set of these secondary interactions occurs when the CMB photons Thomson scatter off bubbles of free electrons, an effect called patchy screening [ 86 , 22 , 56 , e.g., ] . This scattering acts as a screen on the CMB with some optical depth \mathbf{\tau} , washing out the primary signal and generating new polarization [ 87 ] on large angular scales (see Fig. 1 ). Figure 1: A demonstration of the map-level effects of patchy screening and the kSZ effect. The left panel shows the primary, unmodified CMB. Panel A (center) demonstrates the screening that happens to the primary CMB when a bubble of electrons is simulated in the foreground, located inside the dashed green circle. Note that the CMB fluctuations, both temperature (colors) and polarization (black vectors), are ”washed out” by the Thomson scattering. Panel B (right) demonstrates the shift in the CMB temperature fluctuations that happens when the bubble of electrons is given a positive LOS velocity with respect to the Hubble flow. This leads to a colder (redder) CMB spot in temperature; the polarization is unaffected at this order in scattering. Detections of the all-sky average value of \tau have been made using large-scale polarization with both WMAP [ 21 , \bar{\tau}=0.087\pm 0.017 ,] and Planck [ 5 , \bar{\tau}=0.054\pm 0.007 ,] . The angular fluctuations in the \tau field, \tau(\hat{\mathbf{n}}) , reduce the CMB signal more strongly along lines of sight where the electrons are more dense. Both high-redshift reionization processes ( z\gtrsim 6 ) and low-redshift galaxies ( z\lesssim 3 ) create free electrons that contribute to this signal. The EoR \tau fluctuations, “patchy \tau ,” are expected to have an order of magnitude larger variance than their post-EoR counterpart [ 70 , e.g., Fig. 2 in] . Even so, \tau(\hat{\mathbf{n}}) for all redshifts is more difficult to detect than its average value due to current CMB survey noise levels. Forecasts estimate that the patchy \tau auto-power spectrum will likely be detectable with future CMB surveys [ 22 , 37 ] . Roy et al. [70] forecasted a cross-correlation of patchy \tau with galaxies at post-EoR redshifts to constrain the circumgalactic medium (CGM) around these galaxies, and there is a hint of \tau(\hat{\mathbf{n}}) using this method by Coulton et al. [16] with CMB data from the Atacama Cosmology Telescope (ACT) and Planck in cross-correlation with galaxy data from 0\left<\delta_{e}\delta_{e}\delta_{e}\right> , yielding a measure of the electron bispectrum multiplied by the well-understood statistics of the large-scale cosmic bulk flows. This is the same argument that Doré et al. [20] make and several “projected-fields” studies thereafter follow [ 32 , 25 , 40 , 12 ] . The differences here compared to the projected-fields studies are that we use the electron overdensities instead of galaxy ones, and we are probing EoR redshifts rather than post-EoR redshifts. We evaluate this further in Section VIII.2 . A theoretical expression for C_{L}^{K\tau} , valid in all regimes, is C_{L}^{K\tau}\ =A^{K}_{TT}(L)\int_{0}^{\eta_{\mathrm{CMB}}}\frac{d\eta}{\eta^{2}}\ g^{3}(\eta)\ \mathcal{T}\left(k=\frac{L}{\eta},\eta\right) (9) where A^{K}_{TT}(L) is a normalization factor that we apply to our reconstructed K maps in this work and define in the next section. Because \tau and \Theta_{\mathrm{kSZ}} are sourced by the same free electrons, the cross-correlation between the \tau and K maps is sensitive to g^{3}(\eta) ; for a more general tracer with radial distribution W(\eta) , this factor would instead be W(\eta)g^{2}(\eta) , [ 20 , e.g., ] . \mathcal{T}\left(k,\eta\right) is defined as the triangle power spectrum: \mathcal{T}(k,\eta)\equiv\\ \int\frac{d^{2}\mathbf{q}}{(2\pi)^{2}}\ W^{\mathrm{kSZ}}(\mathbf{q}\eta)\ W^{\mathrm{kSZ}}(|\mathbf{k}+\mathbf{q}|\eta)\ B_{\delta_{e}p_{\hat{\mathbf{n}}}p_{\hat{\mathbf{n}}}}(\mathbf{k},\mathbf{q},-\mathbf{k}-\mathbf{q}), (10) where W^{\mathrm{kSZ}} are the filters referenced within Equation 8 and shown in Figure 5 below. B_{\delta_{e}p_{\hat{\mathbf{n}}}p_{\hat{\mathbf{n}}}}(\mathbf{k},\mathbf{q},-\mathbf{k}-\mathbf{q}) is the “hybrid bispectrum” of the electron density fluctuations \delta_{e} and the LOS component of the electron momentum field p_{\hat{\mathbf{n}}} : \displaystyle\langle\delta_{e}(\mathbf{k}_{1})\ p_{\hat{\mathbf{n}}}(\mathbf{k}_{2})\ p_{\hat{\mathbf{n}}}(\mathbf{k}_{3})\rangle \displaystyle= \displaystyle(2\pi)^{3}\delta_{D}(\mathbf{k}_{1}+\mathbf{k}_{2}+\mathbf{k}_{3}) \displaystyle\ B_{\delta_{e}p_{\hat{\mathbf{n}}}p_{\hat{\mathbf{n}}}}(\mathbf{k}_{1},\mathbf{k}_{2},\mathbf{k}_{3}). (11) The triangle power spectrum arises because taking the square of the kSZ signal at the map level leads to a convolution in the Fourier domain: It represents a compression of the hybrid bispectrum B_{\delta_{e}p_{\hat{\mathbf{n}}}p_{\hat{\mathbf{n}}}} as shown by the choice of triangles with one side-length equal to k . Therefore, this cross-correlation of the K and \tau fields exists and should be non-zero. Furthermore, we expect it to be positive because the constituents trace common electron overdensities. Finally, we predict that this cross-correlation traces both EoR and post-EoR physics, just as the patchy \tau autospectrum and the K autospectrum do [ 70 , 78 ] . We quantify the high- and low-redshift components of C_{L}^{K\tau} using simulations described in Sections IV and VI , respectively. III Field Reconstruction XY f^{\tau}_{XY}(\mathbf{l}_{1},\mathbf{l}_{2}) f^{K}_{XY}(\mathbf{l}_{1},\mathbf{l}_{2}) f^{\phi}_{XY}(\mathbf{l}_{1},\mathbf{l}_{2}) TT C_{l_{1}}^{TT}+C_{l_{2}}^{TT} \sqrt{C_{l_{1}}^{TT,\mathrm{kSZ}}C_{l_{2}}^{TT,\mathrm{kSZ}}} (\mathbf{L}\cdot\mathbf{l}_{1})C_{l_{1}}^{TT}+(\mathbf{L}\cdot\mathbf{l}_{2})C_{l_{2}}^{TT} EB [\tilde{C}_{l_{1}}^{EE}-\tilde{C}_{l_{2}}^{BB}]\sin[2\varphi_{l_{1}l_{2}}] — [(\mathbf{L}\cdot\mathbf{l}_{1})\tilde{C}_{l_{1}}^{EE}-(\mathbf{L}\cdot\mathbf{l}_{2})\tilde{C}_{l_{2}}^{BB}]\sin[2\varphi_{l_{1}l_{2}}] Table 2: The mode-coupling expressions that enter the minimum variance filters for \hat{\tau} , \hat{K} , and \hat{\phi} given CMB signals X and Y , where \varphi_{l_{1}}=\cos^{-1}(\hat{\mathbf{n}}\cdot\hat{l_{1}}) , \varphi_{l_{1}l_{2}}=\varphi_{l_{1}}-\varphi_{l_{2}} , and \mathbf{L}=\mathbf{l}_{1}+\mathbf{l}_{2} . In this work, C_{l}^{TT,\mathrm{kSZ}} is the fiducial theory kSZ power spectrum from AMBER [ 84 ] . In practice, both \tau and K are reconstructed from CMB data using quadratic estimators, similar to those used routinely for lensing reconstruction. Obtaining maps of \tau and K in order to cross-correlate them involves isolating specific modes of the CMB temperature and polarization data and multiplying them together in the map domain. This process up-weights the scales that are dominated by the signals of interest. In this section, we introduce the flat-sky approximations of the quadratic estimators for the K and \tau fields in a general form before defining their different filters [ 80 , as done in, e.g.,] . These general equations are the same as those for the CMB lensing potential, \phi , quadratic estimators [ 35 ] , whose filters we also present in Table 2 for reference. The cross-power spectrum of the Fourier transform of two CMB fields X and Y can be expressed as \langle X(\mathbf{l}_{1})Y(\mathbf{l}_{2})\rangle_{\mathrm{CMB}}=(2\pi)^{2}\ \delta(\mathbf{l}_{1}+\mathbf{l}_{2})\ C_{l}^{XY}, (12) where \mathbf{l}_{1} and \mathbf{l}_{2} are the wavevectors describing the plane waves on the sky on our flat surface. The CMB subscript indicates an ensemble average over multiple realizations of the CMB. Given observed CMB maps X and Y , representing temperature or polarization maps T , E , or B , a quadratic estimator for a given field \Gamma which couples modes \mathbf{l}_{1} and \mathbf{l}_{2}\equiv\mathbf{L}-\mathbf{l}_{1} can be written as \begin{split}&\hat{\Gamma}_{XY}(\mathbf{L})=\\ &A^{\Gamma}_{XY}(\mathbf{L})\int\frac{d^{2}\mathbf{l}_{1}}{(2\pi)^{2}}F^{\Gamma}_{XY}(\mathbf{l}_{1},\mathbf{L}-\mathbf{l}_{1})\left[X(\mathbf{l}_{1})Y(\mathbf{L}-\mathbf{l}_{1})\right],\end{split} (13) where A^{\Gamma}_{XY}(\mathbf{L}) is the normalization function, A_{XY}^{\Gamma}(\mathbf{L})=\left[\int\frac{d^{2}\mathbf{l}_{1}}{(2\pi)^{2}}F^{\Gamma}_{XY}(\mathbf{l}_{1},\mathbf{L}-\mathbf{l}_{1})f^{\Gamma}_{XY}(\mathbf{l}_{1},\mathbf{L}-\mathbf{l}_{1})\right]^{-1}, (14) and F^{\Gamma}_{XY}(\mathbf{l}_{1},\mathbf{L}-\mathbf{l}_{1}) is the filtering function, F^{\Gamma}_{XY}(\mathbf{l}_{1},\mathbf{L}-\mathbf{l}_{1})=\frac{f^{\Gamma}_{XY}(\mathbf{l}_{1},\mathbf{L}-\mathbf{l}_{1})}{\alpha C_{l_{1}}^{XX,\mathrm{t}}C_{l_{2}}^{YY,\mathrm{t}}}, (15) with C_{l}^{XX,\mathrm{t}}=C_{l}^{XX}+C_{l}^{XX,\ \text{noise}} , the power spectrum of the total sky emission, including lensed CMB, foregrounds, and instrumental noise. \alpha=1 or 2 depending on the symmetry properties of the estimator; for XY=TT , EE , or BB , \alpha=2 [ 35 , 80 ] . A^{\Gamma}_{XY}(\mathbf{L}) normalizes the estimate \hat{\Gamma} such that it is, on average, equal to the actual \Gamma field. The mode-coupling functions f^{\Gamma}_{XY}(\mathbf{l}_{1},\mathbf{L}) are given in Table 2 for the \tau , K , and \phi mode coupling choices. When assuming “ideal” filtering, meaning that the total power spectra in the filter match the total power in the maps, the noise power spectrum for a given mode-coupling becomes equal to the normalization; i.e., N^{\Gamma}_{XY}(\mathbf{L})=A^{\Gamma}_{XY}(\mathbf{L}) . III.1 \tau Reconstruction & Normalization Patchy optical depth \tau modulates CMB maps according to Equations 4 and 5 , causing new correlations between CMB modes \mathbf{l}_{1} and \mathbf{l}_{2} according to: \langle X(\mathbf{l}_{1})Y(\mathbf{l}_{2})\rangle_{\mathrm{CMB}}\ =f^{\tau}_{XY}(\mathbf{l}_{1},\mathbf{l}_{2})\ \tau(\mathbf{l}_{1}+\mathbf{l}_{2}), (16) with an assumed-fixed field \tau and f^{\tau}_{XY}(\mathbf{l}_{1},\mathbf{l}_{2}) given in Table 2 . See the bottom two panels in Fig. 5 for the specific E and B filters used here for \tau . One can reconstruct this patchy \tau signal in Fourier space, \hat{\tau}(\mathbf{L}) , at a scale \mathbf{L}\neq 0 using the quadratic estimator derived in Dvorkin and Smith [22] , whose flat sky representation is presented in Su et al. [80] This has been applied to real data in e.g. Gluscevic et al. [28] and Namikawa [55] , and is equivalent to Equation 13 with \Gamma=\tau . III.2 K Reconstruction & Normalization The K(\mathbf{L}) field is reconstructed in a similar manner to \tau(\mathbf{L}) and \phi(\mathbf{L}) , by way of a quadratic estimator which filters and multiplies two CMB maps together. In this case, K(\mathbf{L}) is only measurable with a quadratic estimator of the CMB temperature field. This is because the kSZ effect only traces the electron bulk flows, and is not dependent on the incident photons’ polarization. Reconstructed in the manner laid out in Section III , \hat{K}(\mathbf{L}) traces fluctuations in the (squared) velocity field on large scales [ 78 ] but is equivalent, up to potentially the final normalization step, to the squared kSZ fields constructed for projected-fields studies on smaller scales [ 20 , 32 ] . We use the same symbol K to refer to both regimes. Table 2 contains the minimum variance filter for \hat{K}(\mathbf{L}) reconstruction according to equations 13 , 14 , and 15 for \Gamma=K . See the second panel in Fig. 5 for the filters used to reconstruct the K signal in this work. We discuss the origin of the input kSZ signal to this reconstruction in the next section, and discuss our methods for computing the optimal reconstruction noise (as in Eq. 13 ) for CMB-S4 and CMB-HD in Section V . IV EoR Simulations Figure 2: A 3-panel display of 2^{\circ}\times 2^{\circ} cutouts of the fiducial AMBER maps used in this work. Left: The reionization kSZ map showing the changes to CMB temperature that the kSZ effect makes. Center: The \tau map showing the projected electron density fluctuations centered about \bar{\tau}=0.06 . We have added the low-z value of \bar{\tau}_{z<5}=0.029 to the AMBER map to include the optical depth from redshifts below those simulated by AMBER . Right: In purple, the \hat{K}(\hat{\mathbf{n}}) map reconstructed using the kSZ map on the left and overlaid with colored \tau contours from the center map to display the correlations between the two. To obtain a realistic signal for both patchy \tau and kSZ during the EoR, we use the semi-numerical Abundance Matching Box for the Epoch of Reionization ( AMBER ) code [ 84 , 15 ] . AMBER is used to make full-sky Healpix simulations of different fields in the Universe from the EoR, such as CMB lensing, patchy \tau , and patchy kSZ. This code allows us to specify the reionization history, making it ideal for parameter-space studies and for inferring when reionization occurred. There are five reionization parameters that can be varied within AMBER to obtain different reionization fields: the redshift midpoint z_{\text{mid}} , the 5% – 95% ionized redshift duration \Delta z , the redshift asymmetry A_{z} , the minimum halo mass M_{\text{min}} , and the radiation mean-free path l_{\text{mfp}} . See Trac et al. [84] for more details about these parameters. We simulate maps with the same settings presented in Chen et al. [15] , for which these parameters were varied individually about their central fiducial values [ z_{\text{mid}}=8 , \Delta z=4 , A_{z}=3 , M_{\text{min}}=10^{8}M_{\odot} , l_{\text{mfp}}=3 Mpc/h], creating 11 different reionization histories spanning 53000 . Even on large scales, the \delta_{e} bispectrum scaled appropriately by v^{2}_{\mathrm{rms}} contributes to over half of C_{L}^{K\tau} . This confirms that this cross-correlation almost solely probes the non-Gaussianity of electrons during the EoR. To further investigate the theory of C_{L}^{K\tau} introduced in Section II , we perform a test to check what signals this cross-correlation is sourced by. Based on Equations 9 - 11 , it is reasonable to ask whether C_{L}^{K\tau} is dominated by the electron density fluctuations scaled by v^{2}_{\mathrm{rms}} because it depends on the hybrid bispectrum between the electron density and momentum fields. On small scales, velocities vary slowly across the sky, so within a local patch we can potentially approximate p(\hat{\mathbf{n}})\approx v(\hat{\mathbf{n}})\left(1+\delta_{e}(\hat{\mathbf{n}})\right)\approx v_{\mathrm{loc}}\left(1+\delta_{e}(\hat{\mathbf{n}})\right) for the local value of the velocity. This is typically taken to be the dominant regime for projected-fields kSZ studies, as first assumed by Doré et al. [20] . DeDeo et al. [19] investigated the assumptions from Doré et al. [20] in depth, and verified that this \langle vv\rangle\langle\delta\delta\delta\rangle term in the full \langle p_{\hat{\mathbf{n}}}p_{\hat{\mathbf{n}}}\delta\rangle bispectrum, mirrored in Eq. 11 , is dominant on small scales. Though Patki et al. [60] recently found that some terms besides \langle vv\rangle\langle\delta\delta\delta\rangle are stronger than DeDeo et al. [19] expected, we seek to determine whether this term should contribute the large majority of the C_{L}^{K\tau} signal at the scales we probe in this work. Here we investigate whether C_{L}^{K\tau} from the EoR is also within this regime at the scales where it will be measured with upcoming surveys. We test the approximation by passing the fiducial AMBER \tau(\hat{\mathbf{n}}) map through the K reconstruction pipeline, obtaining a normalized \sim\tau_{\mathrm{small-scale}}^{2} map in place of the usual (\Theta_{\mathrm{kSZ,\ small-scale}}^{\mathrm{filt}})^{2} map. We then cross-correlate this reconstructed map with the original \tau(\hat{\mathbf{n}}) map. This directly yields the electron density bispectrum contribution to our overall C_{L}^{K\tau} calculation, B_{\delta_{e}\delta_{e}\delta_{e}} . We make the ansatz that this allows us to estimate B_{\delta p_{\hat{\mathbf{n}}}p_{\hat{\mathbf{n}}}} in C_{L}^{K\tau} from Equation 10 via: B_{\delta_{e}p_{\hat{\mathbf{n}}}p_{\hat{\mathbf{n}}}}\approx\frac{1}{3}v_{\mathrm{rms}}^{2}T_{0}^{2}B_{\delta_{e}\delta_{e}\delta_{e}}. (19) Because C_{L}^{K\tau} contains information about v_{\mathrm{rms}} , we multiply this projected B_{\delta_{e}\delta_{e}\delta_{e}} by \frac{1}{3}\frac{v^{2}_{\mathrm{rms}}}{c^{2}}T_{0}^{2} to make an apples-to-apples comparison of this bispectrum and the original C_{L}^{K\tau} . The 1/3 scaling comes from the fact that we take only the radial component of the three-dimensional velocity field in this projection, and the scaling by the primary CMB temperature T_{0} corrects for the fact that \tau maps are unitless, while our kSZ maps are in \mu K. Given that the rms value of the bulk-flow velocities is not a strong function of redshift, for this test we replace it by a constant value at the fiducial midpoint of reionization in our fiducial AMBER mocks, z=8 . We use the Code for Anisotropies in the Microwave Background [ 43 , CAMB,] to estimate v^{2}_{\mathrm{rms}}(z=8)\approx 240 km s -1 . The result of this test is shown in Figure 9 and demonstrates that these two bispectra are indeed in good agreement. On small scales, (L>3000) , they match as expected, and on larger scales this electron bispectrum constributes more than half of the total C_{L}^{K\tau} signal. Therefore, our ansatz in Equation 19 is validated. This means that a measurement of C_{L}^{K\tau} in the manner described in this work will directly probe B_{\delta_{e}\delta_{e}\delta_{e}} , something that will yield information solely about the skewness, or bispectrum, of the non-Gaussianity in the \delta_{e} field during reionization. By contrast, the other higher-point functions we studied measure even-numbered statistics of the electron distribution during the EoR. Therefore, C_{L}^{K\tau} contains unique, novel information. IX Discussion We have introduced a new CMB estimator for studying reionization: the cross-correlation between reconstructed maps of patchy screening and the squared kinetic Sunyaev-Zel’dovich fluctuations, C_{L}^{K\tau} . These two types of reconstructed maps have been considered as separate probes of the reionization process; here, we investigated the potential gains from analyzing them together in cross-correlation. We found that the C_{L}^{K\tau} signal is appreciable and positive because both trace the same bubbles of ionized gas during reionization. To determine the detectability of this signal, we considered the case in which both maps are reconstructed from the same CMB survey — using the CMB temperature information for the \hat{K} reconstruction, and, to avoid the leading source of foreground bias, using the CMB polarization information for estimating the patchy \tau noise. We forecasted signal-to-noise ratios for the \tau auto-power spectrum C_{L}^{\tau\tau} , the K auto-power spectrum C_{L}^{KK} , and C_{L}^{K\tau} using expected noise levels for the CMB-S4 and CMB-HD surveys. We found that the outlook for a detection of this signal is promising for future CMB experiments, specifically that CMB-S4 should make a 0.54-1.8\sigma measurement and CMB-HD should make a 12-14\sigma measurement, depending on the amount of power from other extragalactic foregrounds remaining in the kSZ maps after the multi-frequency ILC cleaning process is performed. Even considering the most pessimistic SNR, an upper limit on C_{L}^{K\tau} will yield valuable information about the midpoint and duration of reionization via this new, independent method from the current best constraints. We quantified this by explicitly measuring how the auto- and cross-power spectra depend on the 5 reionization parameters varied in the AMBER mocks, finding significant sensitivity to these properties. Specifically, we found that C_{L}^{K\tau} is a complementary way to measure \Delta z and is uniquely sensitive to the skewness of reionization in redshift, A_{z} , making it a probe of the early phases of reionization. We also investigated how much the low-redshift gaseous halos surrounding galaxies contribute to C_{L}^{K\tau} by using the Agora post-EoR simulations. We found that, after masking the brightest halos, these lower redshifts contribute a roughly equal amount of C_{L}^{K\tau} signal as the EoR does. Even so, this low-redshift component will likely become less theoretically uncertain than it is now with the advent of new kSZ analyses at these low redshifts in the coming years. To avoid the most detrimental source of bias from other non-Gaussian extragalactic sources, which appear in the temperature four-point function, we conservatively focused on the polarization estimator for reconstructing the \tau maps. However, it is possible that these terms could be brought under control using tools developed for CMB lensing analyses. If we instead use both the temperature and polarization estimators for the \tau forecast, the signal-to-noise on C_{L}^{K\tau} increases by \sim 25-30 % and that of C_{L}^{\tau\tau} by a factor of \sim 5 times. We found that C_{L}^{K\tau} is mainly sourced by small-scale angular fluctuations in the optical depth and their cross-correlation with small-scale squared optical depth fluctuations. This makes it the high-redshift analog of the projected-fields method of measuring the kSZ signal. In this view, the C_{L}^{K\tau} spectrum is directly probing the bispectrum, or spatial skewness, of the ionized bubbles during reionization. This makes it a distinct probe from both C_{L}^{KK} and C_{L}^{\tau\tau} , which measure other aspects of the morphology of the reionization process. In summary, we found that the new cross-correlation statistic we isolated will be a new and complementary way to constrain the epoch of reionization using upcoming CMB surveys. This further establishes the degree to which these surveys will be critical tools for uncovering the elusive nature of the earliest generations of stars and galaxies.