1: \documentclass{emulateapj}
2: % \documentclass[preprint,12pt]{aastex}
3:
4: \bibliographystyle{apj}
5: \usepackage{amsmath}
6: \usepackage{amssymb}
7: \usepackage{epstopdf}
8:
9: % Formatting
10: \renewcommand{\vec}{\boldsymbol}
11: \newcommand{\dd}[2]{\ensuremath{ \frac{ \partial #1 }{ \partial #2 } }}
12: \newcommand{\ddd}[2]{\ensuremath{ \frac{ \partial^2 #1 }{ \partial #2^2 } }}
13: \newcommand{\grad}{\nabla}
14: \newcommand{\de}{\mathrm{d}}
15:
16: \newcommand{\mc}{\multicolumn{2}{c}}
17: \newcommand{\mcf}{\multicolumn{4}{c}}
18: \newcommand{\numcol}{r@{.}l}
19: \newcommand{\head}{\multicolumn{1}{c}}
20:
21: % Variables
22: \newcommand{\DM}{\ensuremath{\mathrm{DM}}}
23: \newcommand{\n}{\ensuremath{n_{\mathrm{e}}}}
24: \newcommand{\eikz}{e^{i (k z - \omega t)}}
25: \newcommand{\xhat}{\ensuremath{\vec{\hat{x}}}}
26: \newcommand{\yhat}{\ensuremath{\vec{\hat{y}}}}
27: \newcommand{\zhat}{\ensuremath{\vec{\hat{z}}}}
28: \newcommand{\thscatt}{\ensuremath{\theta_{\mathrm{scatt}}}}
29: \newcommand{\bt}{\ensuremath{\vec{\theta}}}
30: \newcommand{\veff}{\ensuremath{V_{\mathrm{eff}}}}
31: \newcommand{\veffa}{\ensuremath{V_{\mathrm{eff},\alpha}}}
32: \newcommand{\veffd}{\ensuremath{V_{\mathrm{eff},\delta}}}
33: \newcommand{\vp}{\ensuremath{V_{\mathrm{p}}}}
34: \newcommand{\vpa}{\ensuremath{V_{\mathrm{p},\alpha}}}
35: \newcommand{\vpd}{\ensuremath{V_{\mathrm{p},\delta}}}
36: \newcommand{\vobs}{\ensuremath{V_{\mathrm{obs}}}}
37: \newcommand{\vobsa}{\ensuremath{V_{\mathrm{obs},\alpha}}}
38: \newcommand{\vobsd}{\ensuremath{V_{\mathrm{obs},\delta}}}
39: \newcommand{\vscreen}{\ensuremath{V_{\mathrm{screen}}}}
40: \newcommand{\vveff}{\ensuremath{\vec{V}_{\mathrm{eff}}}}
41: \newcommand{\vveffhat}{\ensuremath{\vec{\hat{V}}_{\mathrm{eff}}}}
42: \newcommand{\vvp}{\ensuremath{\vec{V}_{\mathrm{p}}}}
43: \newcommand{\vvobs}{\ensuremath{\vec{V}_{\mathrm{obs}}}}
44: \newcommand{\vvscreen}{\ensuremath{\vec{V}_{\mathrm{s}}}}
45: \newcommand{\len}{\ensuremath{\mathcal{L}}}
46: \newcommand{\scriptA}{\ensuremath{\mathcal{A}}}
47: \newcommand{\scriptB}{\ensuremath{\mathcal{B}}}
48: \newcommand{\rlc}{\ensuremath{r_{lc}}}
49: \newcommand{\scr}{\ensuremath{s}}
50:
51: \newcommand{\scrfrac}{\ensuremath{ \left(\frac{1 - \scr}{\scr} \right) }}
52: \newcommand{\scrfracinv}{\ensuremath{ \left(\frac{\scr}{1 - \scr} \right) }}
53:
54: % Units
55: \newcommand{\cucm}{\ensuremath{\textrm{ cm}^{-3}}}
56: \newcommand{\pc}{\ensuremath{\textrm{ pc}}}
57: \newcommand{\kpc}{\ensuremath{\textrm{ kpc}}}
58: \newcommand{\AU}{\ensuremath{\textrm{ AU}}}
59: \newcommand{\cm}{\ensuremath{\textrm{ cm}}}
60: \newcommand{\meter}{\ensuremath{\textrm{ m}}}
61: \newcommand{\marcsec}{\ensuremath{\textrm{ mas}}}
62: \newcommand{\uarcsec}{\ensuremath{\textrm{ } \mu \textrm{as}}}
63: \newcommand{\mas}{\ensuremath{\textrm{ mas}}}
64: \newcommand{\uas}{\ensuremath{\textrm{ } \mu \textrm{as}}}
65: \newcommand{\s}{\ensuremath{\textrm{ s}}}
66: \newcommand{\rad}{\ensuremath{\textrm{ rad}}}
67: \newcommand{\us}{\ensuremath{\, \mu \textrm{s}}}
68: \newcommand{\dy}{\ensuremath{\textrm{ day}}}
69: \newcommand{\yr}{\ensuremath{\textrm{ yr}}}
70: \newcommand{\MHz}{\ensuremath{\textrm{ MHz}}}
71: \newcommand{\GHz}{\ensuremath{\textrm{ GHz}}}
72: \newcommand{\Hz}{\ensuremath{\textrm{ Hz}}}
73: \newcommand{\kHz}{\ensuremath{\textrm{ kHz}}}
74: \newcommand{\mHz}{\ensuremath{\textrm{ mHz}}}
75: \newcommand{\Msun}{\ensuremath{\, M_{\sun}}}
76: \newcommand{\kg}{\ensuremath{\textrm{ kg}}}
77: \newcommand{\g}{\ensuremath{\textrm{ g}}}
78: \newcommand{\km}{\ensuremath{\textrm{ km}}}
79: \newcommand{\m}{\ensuremath{\textrm{ m}}}
80: \newcommand{\kms}{\ensuremath{\textrm{ km s}^{-1}}}
81: \newcommand{\masyr}{\ensuremath{\textrm{ mas yr}^{-1}}}
82: \newcommand{\dB}{\ensuremath{\textrm{ dB}}}
83: \newcommand{\J}{\ensuremath{\textrm{ J}}}
84: \newcommand{\K}{\ensuremath{\textrm{ K}}}
85:
86: \newcommand{\fnu}{\ensuremath{f_\nu}}
87: \newcommand{\ft}{\ensuremath{f_t}}
88: \newcommand{\barne}{\ensuremath{\bar{n}_e}}
89:
90: \begin{document}
91:
92: \title{Deflection of pulsar signal reveals compact structures in the Galaxy}
93:
94: \shorttitle{Deflection of Pulsar Signal Reveals Structures}
95: \shortauthors{Hill et al.}
96:
97: \author{Alex S.~Hill, Daniel R.~Stinebring, Curtis T.~Asplund, Daniel~E.~Berwick, \\
98: Wendeline~B.~Everett \& Natalie R.~Hinkel}
99:
100: \affil{Oberlin College, Department of Physics and Astronomy, Oberlin, OH 44074, USA}
101:
102: \begin{abstract}
103: We have detected a strong deflection of radio waves from the pulsar
104: PSR B0834+06 in scintillation observations. Interference between the undeflected
105: pulsar image and deflected subimages allows single dish interferometry of the
106: interstellar medium with sub-milliarcsecond resolution. We infer the presence of
107: scattering structure(s) similar to those that are thought to cause Extreme
108: Scattering Events in quasar flux monitoring programs: size $\sim 0.2 \AU$
109: (an angular size of $0.1 \mas$) with an electron overdensity of $\gtrsim 10^3$
110: compared to the warm ionized medium. The deflectors are nearly stationary in a
111: scattering screen that is thin ($\lesssim 5 \%$ of the pulsar-observer distance
112: in extent), is located $70 \%$ of the way from the Earth to the pulsar, and
113: has been seen consistently in observations dating back $20$ years. The pulsar
114: scans the scattering screen at a velocity of $110 \kms$ with a detection radius
115: of $15 \mas$. Pulsar observations such as these --- particularly with a new
116: generation of low-frequency radio telescopes with large collecting areas ---
117: hold promise for improving constraints on the poorly understood physical
118: characteristics and space density of the deflecting structures.
119: Such observations may also
120: prove useful in correcting deviations the deflectors produce
121: in high-precision timing of millisecond pulsars.
122: \end{abstract}
123:
124: \keywords{ISM: structure --- pulsars: general --- scattering --- techniques:
125: spectroscopic --- pulsars: individual (PSR~B0834+06)}
126:
127: \section{Introduction}
128: The discovery of Extreme Scattering Events \citep[ESEs;][]{fdj+87} in the late 1980s provided evidence for ultracompact
129: ($\lesssim 1 \AU$), ionized objects in the interstellar medium. Despite
130: more than 15~years of observational and interpretive work, the physical
131: conditions, origin, and lifetime of these compact refractors are not well
132: established \citep{rbc87,f94,rlg97,ww98}. Furthermore, their space density and
133: possible contribution
134: to the mass distribution in the Galaxy are poorly known. The basic difficulty
135: has been detecting events in either quasar or pulsar monitoring
136: programs \citep{w01}. In order to substantially affect the flux density
137: of a radio source, the refractor must be nearly aligned with the source.
138: The refractors are typically located at distances of kiloparsecs, so a
139: milliarcsecond alignment is required. Thus, only about a dozen ESEs
140: have been identified, and few of these are easy to model.
141:
142: The requirement of stringent alignment is relaxed if the intrinsic source size
143: is small enough to exhibit interstellar scintillation, one of the
144: observational consequences of coherent multipath scattering from density
145: inhomogeneities in the medium. Pulsars are the dominant source in this category.
146: Interference between various parts of the pulsar image gives rise to faint but
147: observable periodic fringing patterns in pulsar dynamic spectra (radio flux
148: density as a function of time and observing frequency). These features have a
149: clear representation in the secondary spectrum (the squared modulus of the
150: Fourier transform of the dynamic spectrum) and yield information about the
151: location of the scattering material and the structure of the image \citep{ra82,
152: hwg85,cw86, wc87, rlg97}. In particular, periodic fringing in the dynamic
153: fringing gives rise to parabolic \emph{scintillation arcs} \citep{s01,h03,
154: crs+04, wms+04} in the secondary spectrum, which imply that the dominant
155: scattering material is localized in a thin screen along the line of sight.
156:
157: In this Letter, we present evidence for multiple imaging of the pulsar
158: PSR~B0834+06 with greater detail than a similar event reported by \citet{rlg97}.
159: We detected secondary spectrum features caused by at least five separate ray
160: paths throughout a $26$ day observing run. The angular separation
161: between the undeflected pulsar image and each of the four deflected subimages
162: grew linearly at a rate consistent with the velocity of the pulsar. This,
163: combined with multi-frequency data, imply that the multiple imaging is caused by
164: one or several deflecting structures similar to those that cause ESEs and that
165: the structures are stationary in the scattering screen. Such objects, if they
166: trace out underlying neutral material \citep{ww98}, may contain a significant
167: fraction of the mass of the Galaxy.
168:
169: \section{Thin Screen Model}
170:
171: We use a model of scintillation arcs developed in \citet{s01}, \citet{h03},
172: \citet{crs+04}, and \citet{ wms+04} and summarized here. The essential feature
173: of the model is that a scintillation arc is caused by scattering in a thin
174: screen along the line of sight to the pulsar. Interference between a bright core
175: of the pulsar image and the scatter-broadened halo gives rise to the basic arc,
176: and the details of the power distribution along the arc provide information
177: about the power distribution in the pulsar image. Therefore, our secondary
178: spectra of PSR~B0834+06 at $327 \MHz$ provide a partial image of the pulsar with
179: an angular resolution of $\sim 0.3 \mas$ and a field of view of $\sim 40 \mas$ (see below).
180:
181: \subsection{Image features and interference patterns}
182:
183: We define a coordinate system in the plane of the sky with the pulsar at the
184: origin, the $\bt_x$-axis pointing along the effective velocity
185: vector, and $\bt_y \perp \bt_x$. The pulsar effective velocity \citep{cr98},
186: \begin{equation} \label{eq:veff}
187: \vveff = (1 - \scr) \vvp + \scr \vvobs - \vvscreen,
188: \end{equation}
189: is the apparent velocity of the image of the pulsar through the thin scattering
190: screen, where \scr\ is the fractional position of the screen from
191: the pulsar ($\scr = 0$) to the observer ($\scr = 1$), \vvp\ is the transverse
192: velocity of the pulsar, \vvobs\ is the transverse velocity of the observer, and
193: \vvscreen\ is the transverse velocity of the screen. Interference between two
194: arbitrary points in the image plane, $\bt_1$ and $\bt_2$, causes fringing in
195: the dynamic spectrum with conjugate time and conjugate frequency of
196: $\ft = - (\bt_2 - \bt_1) \cdot \vveff/(\scr \lambda)$ and
197: $\fnu =D (1-\scr) (\bt_2^2-\bt_1^2)/(2 \scr c)$,
198: where $\lambda$ is the observing wavelength. The $\fnu$ coordinate is a measure
199: of differential time delay between pairs of rays, and $\ft$ represents the
200: temporal fringe frequency of the interference between the two rays or,
201: alternatively, the differential Doppler shift between them. Interference between
202: the ray at the origin and points along the $\theta_x$-axis produces a parabolic
203: scintillation arc defined by $f_\nu = \eta f_t^2$, where the arc curvature
204: parameter is
205: \begin{equation} \label{eq:eta}
206: \eta = \frac{D \lambda^2 \scr (1 - \scr)}{2 c \veff^2}.
207: \end{equation}
208: Interference between the origin
209: and points with non-zero $\theta_y$ places power inside the parabola
210: because $\fnu \propto \theta_x^2 + \theta_y^2$.
211: Interference between a bright spot in the
212: periphery of the image and the the rest of the image produces an inverted
213: parabola or {\em arclet} with the same $| \eta |$ and a vertex with an $\ft$
214: coordinate that is related to the $\theta_x$-coordinate\footnote{The
215: minus sign in equation~\ref{eq:thetax} and in the defintion of $f_t$ is present because negative $f_t$ values correspond to deflected rays in front of the moving pulsar; points behind the moving pulsar (past closest encounter to the deflecting structure) produce a positive $f_t$ feature.}
216: of the bright spot by
217: \begin{equation} \label{eq:thetax}
218: \theta_{x} = - \left( \frac{\scr \lambda}{\veff}\right) f_t.
219: \end{equation}
220: Such features are accentuated when the image is elongated along the velocity
221: vector or the bright spot lies near the $\theta_x$-axis, or both.
222:
223: \subsection{Screen Location}
224: For pulsars with measured proper motions and measured or estimated distances, we
225: can determine the location of the dominant scattering material from the
226: curvature of the main scintillation arc using equation (\ref{eq:eta}).
227: PSR~B0834+06 has an estimated distance \citep{cl03} and a moderately well
228: determined proper motion \citep{las82}: $D = 0.64 \pm 0.08 \kpc$ and
229: $\mu = 51 \pm 3 \mas \yr^{-1}$. We have measurements of scintillation arcs in
230: PSR~B0834+06 dating back to 1981 (Stinebring et al. 2005, in preparation).
231: Those data and the observations presented here are consistent with a value of
232: $\eta = 0.47 \pm 0.03~{\rm s}^3$ at $327 \MHz$. This results in a value of
233: $\scr = 0.29 \pm 0.04$ or a screen that is located a distance
234: $0.46 \pm 0.08 \kpc$ from the observer.
235: These parameters result in a conversion between $f_t$ and angle on the sky
236: of 50~mHz = 24~mas. Hence, the angular resolution of the secondary spectrum is
237: (2 x 24~mas / 180 pixels) = 0.27~mas per pixel; the useful field of view is about 40~mas.
238:
239: \section{Observations}
240:
241: To search for evidence of compact refractors using pulsar scintillation, we
242: observed the pulsar PSR~B0834+06 at the Arecibo Observatory in 2004 January
243: using the same technique as \citet{h03}. An example dynamic spectrum is shown in
244: Figure~\ref{fig:0834jandyn}. We obtained dynamic spectra during 30--60 minute
245: integrations using the $327 \MHz$ receiver with spectral resolutions of
246: approximately $1.5 \kHz$ on 11 days over a 26 day period. Typically, we
247: simultaneously took data with center frequencies of $321$ and $334 \MHz$ using
248: multiple Wideband Arecibo Pulsar Processor spectrometers. The
249: secondary spectrum in Figure~\ref{fig:0834jansec} exhibits a scintillation arc
250: as well as numerous arclets. The four isolated arclets, labeled $a$--$d$,
251: shifted upward and to the right along the main parabola during the month. This
252: had never been seen before and is the central observational result of this
253: paper.
254:
255: \begin{figure}
256: \plotone{f1}
257: \caption{The dynamic spectrum of PSR B0834+06 observed on
258: 2003 Dec 31. The flux density as a function of frequency and time is shown
259: using a grayscale that is linear in power, with dark regions indicating high
260: power. The crisscross pattern is due to radio waves reaching the observer
261: from a variety of angles ($\sim$~10~mas away from the pulsar position), as
262: detailed in the text.}
263: \label{fig:0834jandyn}
264: \end{figure}
265:
266:
267: \begin{figure}
268: \plotone{f2}
269: \caption{The secondary (or delay--fringe-frequency)
270: spectrum corresponding to the dynamic spectrum in
271: Figure~\ref{fig:0834jandyn}. The grayscale is logarithmic in power and
272: represents a total range of about $10^5$ from the maximum power to the noise
273: level. A primary arc with curvature $\eta = 0.47 \s^3$, fitted to the left side
274: of the main parabola, is shown with a dashed line. Numerous inverted parabolas,
275: or \emph{arclets}, are present with vertices near the primary arc.
276: The four isolated arclets (labeled $a$--$d$) each correspond to a
277: distinct enhancement in the pulsar image. The horizontal axis can be converted
278: into a separation angle projected along the direction of pulsar motion using
279: $2 \mHz \approx 1 \mas$.}
280: \label{fig:0834jansec}
281: \end{figure}
282:
283: \section{Motion of Arclets}
284:
285: We developed an algorithm to locate the position of these arclets in secondary
286: spectra by summing along parabolic regions with vertices incremented along the
287: main parabola. In Figure~\ref{fig:arclet_freq}, we plot the fringe frequency
288: of arclet $a$ as a function of day number for two closely spaced
289: frequencies.
290: Two models of the refracting structures predict different frequency
291: behavior: if the rays are refracted by smoothly varying electron density
292: variations , the angular position of the rays would scale as is
293: typical for the interstellar medium: $\theta \propto \lambda^2$
294: \citep[e. g.][]{s68}; then, $f_t \propto \theta \lambda^{-1} \propto \lambda$.
295: Alternatively,
296: if the rays giving rise to the arclets pass through a discrete and
297: dense refracting structure (lens model) \citep{cfl98}, they would arrive from nearly the same
298: angle $\theta$ at different frequencies; they still deflect as $\lambda^2$ within the structure, but only an unmeasurably slight deviation with wavelength occurs
299: because of the small size and high density of the refracting structure.
300: Thus,
301: $f_t \propto \theta \lambda^{-1} \propto \lambda^{-1}$.
302: We performed a least-squares linear fit to the lower frequency data. We then
303: calculated lines with the same slope but vertical offsets scaling as in the
304: lens model and in the smooth variation model. The data in
305: Figure~\ref{fig:arclet_freq} clearly support the lens model, which we assume in
306: the following discussion.
307:
308: \begin{figure}
309: \plotone{f3}
310: \caption{The fringe frequency ($f_t$)
311: coordinate of the vertex of arclet $a$ versus day number (day 0 = 2003 Dec 31).
312: Data taken at $321 \MHz$ are plotted with filled circles; $334 \MHz$ data are
313: plotted with open circles, and the uncertainty in position is about the diameter
314: of the symbols. The solid line represents a two parameter fit to the $321 \MHz$
315: data. The dashed line has the same slope, but with the offset expected if
316: $f_t \propto \lambda^{-1}$, or $\theta$ is constant with observing wavelength;
317: the dotted line has the offset expected if $f_t \propto \lambda$, or scattering
318: angle $\theta \propto \lambda^2$. These data indicate that a lens-like
319: structure or structures in the scattering screen causes the arclets.}
320: \label{fig:arclet_freq}
321: \end{figure}
322:
323: Using measured values of $f_t$ for the vertices of the arclets, we calculated
324: their angular position as a function of time, as shown in
325: Figure~\ref{fig:arclet_pos}. Uncertainties in the angular positions were
326: typically $0.2 \mas$ as derived from the scatter about the best fit lines.
327: Arclet $a$ moved at a rate of $47 \pm 2 \masyr$ and arclets $b$--$d$ each
328: moved at $51 \pm 2 \masyr$ along the pulsar effective velocity vector.
329: Taking the measured proper motion of the pulsar and accounting for the Earth's
330: motion at this epoch \citep{kamp,cr98} yields an effective pulsar angular
331: velocity of $\mu_{\rm eff} = 50 \pm 3 \masyr$. Hence the angular motion of the
332: pulsar, determined from astrometry, and the rate of increase of the
333: pulsar--deflector angle, determined from the data reported here, are consistent
334: within their combined uncertainties. We use this comparison to estimate the
335: motion of the deflector relative to the screen. We find
336: $v_{\rm scr} = (1-\scr) D (\mu_\mathrm{arc}-\mu_\mathrm{eff}) \lesssim 9 \kms$,
337: implying that the physical objects giving rise to the arclets are essentially
338: stationary in the scattering screen.
339:
340: \begin{figure}
341: \plotone{f4}
342: \caption{The angular position of four arclets over 26
343: days of observations at two observing frequencies (day 0 = 2003 Dec 31). As in
344: Figure~\ref{fig:arclet_freq}, filled circles denote $321 \MHz$ data and open
345: circles indicate $334 \MHz$ data. The uniform, consistent
346: motion of the arclets is primarily due to the motion of the pulsar, indicating
347: that arclets are caused by scattering objects essentially stationary in the
348: screen.}
349: \label{fig:arclet_pos}
350: \end{figure}
351:
352: \section{Discussion}
353: The refracting angle of a plasma lens of size $a$ is $\theta_r \sim \alpha
354: a/D$, where $\alpha \equiv \lambda^2 r_e N_0 D/ \pi a^2$ characterizes the
355: strength of the refractive effects, $N_0 = \int n_e ds$ is the maximum electron
356: column density through the lens, and $r_e$ is the classical electron radius
357: \citep{cfl98}. If we define an average electron density $\barne \equiv N_0 / a$,
358: then $\barne = (5.4 \cucm) \theta_{\rm r}/\lambda^2$, where the angle is
359: measured in mas and the wavelength is expressed in meters. For
360: $\theta_r \approx 15 \mas$ at $327 \MHz$, we find $\barne \approx 100 \cucm$,
361: which is more than $10^3$ times greater than the average electron density in the
362: interstellar medium \citep{cl02}.
363:
364: The thinness of the arclets is remarkable as well and reinforces the interpretation of refraction by lens-like structures embedded in a thin scattering screen.
365: The arclets in Figure~\ref{fig:0834jansec} have a fractional thickness of
366: $\Delta \fnu / \fnu \approx$~5\%. Since $\fnu \propto \theta^2$,
367: this implies
368: $a \sim \Delta\theta (1-\scr) D \approx$~0.1~AU if each arclet
369: is produced by a separate refractor. In this case the mass in each refractor is
370: only $M \sim a^3 \barne m_H \approx 10^{-18} \Msun$; here $m_H$ is the mass of
371: a hydrogen atom and we are assuming a fully ionized object.
372:
373: Instead of four separate deflecting structures, the four arclets may be caused
374: by two or even just one structure.
375: The basic imaging features of a plasma lens become apparent by analyzing a
376: simple model with a Gaussian column density \citep{cfl98}. This model predicts
377: caustic surfaces where rays intersect and a defocusing region directly in front
378: of the lens. In the region between the inner caustic and the outer caustic an
379: observer would see a bright undeflected image of the source and two dim,
380: deflected subimages, which would give rise to two arclets as the subimages
381: interfere with the central image. The angular separation between a pair of
382: arclets remains constant, as in these observations, for the arclet pairs $ab$
383: and $cd$, if the refractive strength parameter $\alpha \gg 1$; in this case,
384: $\Delta \theta \approx 2a / (D - \scr D)$ and
385: $\barne = \pi \alpha (1 - \scr) \Delta \theta / (2 \lambda^2 r_e) \approx
386: (1.9 \cucm) \alpha \Delta \theta_\mathrm{mas} / \lambda^2_\mathrm{meter}$.
387: We can estimate $\alpha$ from the observations because the brightness of the
388: deflected image relative to the central image is $\approx (1 + \alpha)^{-1}$.
389: The peak power of the arclets in Figure~\ref{fig:0834jansec} is about $10^{-3}$
390: of the power at the origin in the secondary spectrum, so $\alpha \gtrsim 10^3$.
391: Using this model, our observations yield estimates of $a \approx 0.25 \AU$,
392: $\barne \sim 2000 \cucm$, and $M \approx 4 \times 10^{-16} \Msun$.
393: It should be noted that neither of these models is in pressure balance in the
394: warm ionized phase of the interstellar medium; hence, they
395: would be expected to dissipate quickly unless constrained through some other
396: mechanism \citep{rbc87, rlg97, ww98}.
397:
398: \citet{ww98} proposed a model for ESEs in which compact ($\sim 1 \AU$),
399: long-lived spheroids of neutral hydrogen are surrounded by a photoionized shell.
400: This results in a double-peaked electron column density profile; each peak may
401: give rise to two enhancements in the image, resulting in the observed
402: four-arclet pattern. In this case,
403: $a \approx \Delta\theta_{ad} (1 - \scr) D/2 \approx 1 \AU$,
404: where $\Delta\theta_{ad} = 3.3 \mas$ is the separation between arclets $a$ and
405: $d$. The mass encompassed by the refracting structure is dramatically higher in
406: this model ($M \sim 10^{-3} M_\odot$) because of the unionized, cold, dense
407: core. A Galactic halo population of these clouds, sufficient to account for the
408: poorly constrained ESE rate, would be a major contributor to the total mass of
409: the Galaxy.
410:
411: Because interference effects are present even when the deflector does not
412: intercept the line of sight, pulsar scintillation observations will
413: improve constraints on the space density of the deflecting structures.
414: The angular offset of the deflector(s) from the $\theta_x$-axis is an important
415: parameter in estimating their space density. Analysis of these data yields
416: $\theta_y$ offsets in the range $2$ -- $6 \mas$.
417: Our sensitivity to this level of offset, along with
418: the ability to detect deflectors within $10$--$15 \mas$ on either side of the pulsar,
419: greatly increases the detection probability compared to flux-only observations.
420:
421: \section{Conclusions}
422:
423: Scintillation observations provide greater detail on the physical properties of
424: these refractors than quasar flux monitoring for two principal reasons. First,
425: the interference between the deflected and undeflected images of the pulsar
426: allows a single dish telescope to act as an interferometer with excellent
427: angular resolution. Second, pulsars scan the scattering screen quickly because
428: of their high transverse velocity. The data presented here demonstrate that
429: dense, compact refractors similar to those that give rise to ESEs can be
430: detected in pulsar scintillation observations.
431: The refracting structures are of interest in their own right, but they also
432: affect high-precision timing of pulsars \citep{cog93}. Since this is one of the most
433: promising methods of detecting a background of primordial gravitational
434: waves, the ability to detect -- and potentially correct for -- the effects of
435: refracting structures is of practical importance as well.
436:
437: We currently do not have enough data containing arclets to decide
438: between models or to place realistic limits on the space density of refracting
439: structures. Arclets are not rare, however, particularly at lower frequency
440: ($\lesssim 400 \MHz$) where the effects of scattering and refraction are more
441: pronounced. We have seen arclets, on occasion, in data from pulsars (PSR)
442: B0355+54, B0823+26, B0834+06, B1133+16, B1642--03, B1737+13, and B1919+21.
443: \citet{rlg97} identified what is probably a poorly-resolved arclet in B0834+06,
444: and \citet{cw86} and \citet{wc87} report similar features in observations of
445: B0919+06 and B1237+25. We are analyzing all of these data further in
446: order to obtain a realistic estimate of the space density of deflectors. It is
447: clear, however, that single-dish scintillation observations with adequate
448: sensitivity at low frequency can be used to track deflectors with milliarcsecond
449: resolution and to estimate their physical properties.
450:
451: \acknowledgments
452: The observations were made at the Arecibo Observatory, which is operated
453: by Cornell University under a cooperative agreement with the National Science
454: Foundation. The research was supported by National Science Foundation grant
455: AST 00-98561.
456:
457: \bibliography{references}
458:
459: \end{document}
460:
461: