Elasticsearch slow extract data - performance

Question

I'm looking for the best method to export some datas from elasticsearch.
My index structure is very simple : ID + 4/5 simple fields
The ID is the ID of French companies,
The other fields are, for example, the activity code, the legal form code, employee size, zip code, ...
The total of records is about 30M.
The count is very fast, but if I want to extract only the IDs, it take a long time.
I need this list "online" for other treatments in a webpage.
How can I do this ?
My query for search is :
{"query":{"bool":{"must":[{"bool":{"should":[{"match":
{"etatadministratifunitelegale":"A"}}]}},{"bool":{"should":[{"match":
{"etatadministratifetablissement":"A"}}]}},{"bool":{"should":[{"match":
{"activiteprincipaleunitelegale":"68.20A"}}]}}], "must_not":[], "should":[] } } }
It's $params in my php code.
This query gives more than 650 000 IDs. To extract them :
$hosts = [
'http://xxxxxxxxx:9200'
];
$client = Elasticsearch\ClientBuilder::create()
->setHosts($hosts)
->build();
$xparams['index'] = 'vecteur';
$xparams['type'] = '_doc';
$xparams['stored_fields'] = [];
$xparams['scroll'] = '1s';
$xparams['size'] = '50000';
$xparams['body'] = $params;
$res = $client->search($xparams);
$scroll_id = $res['_scroll_id'];
$zzz = $res['hits']['hits'];
$xstr = "";
foreach ($zzz as $elt) {
$xstr .= $elt['_id']."\n";
}
fwrite($myfile, $xstr);
while (\true) {
$params2['scroll'] = '1s';
$params2['scroll_id'] = $scroll_id;
$response = $client->scroll($params2);
$zzz = $response['hits']['hits'];
$xstr = "";
foreach ($zzz as $elt) {
$xstr .= $elt['_id']."\n";
}
fwrite($myfile, $xstr);
if (count($response['hits']['hits']) > 0) {
$scroll_id = $response['_scroll_id'];
} else {
break;
}
}
fclose($myfile);
$myfile is a CSV file. Is it clear ?